Ha, ha! I worked on 14-inch Winchester drives in the early 1980s. The unit shown in this video isn't the complete drive; it's just the head-disk assembly (HDA). The full drive, with power supplies and electronics cards was about the size of a washing machine, and that still wasn't enough to get the data on and off the discs, as the drives had to be connected to a controller unit that handled a string of 8-16 drives. This drive is more advanced, having an integral spindle motor, but the 14-inch drives that preceded it in the 1980s were belt-driven by 3-phase induction motors about the size of what you would find in a washing machine. The systems needed an air-conditioned room in order to work reliably. It's amazing that today's hard drives have all of that crammed into a pocket-size package with orders of magnitude higher reliability and lower power consumption!
Thank you so much for commenting. I knew I wasn't crazy. In 1998 I had to go on site to some railyard at night for a quote-unquote hard drive replacement. Arriving in a room with no lights I have to use my flashlight. I open a crate and I swear among all the padding I saw a giant belt-driven something that looks like a sewing machine. I slowly closed the box and backed away. Pretty weird when you're expecting to find something the size of a VHS tape or smaller. I actually found this video looking for what that was. I'm looking up Legacy hard drives Etc. If you have any Google image links or machine type descriptions for what you think that was I saw or something related, please reply. Again thanks.
You're thinking of the 3330 and 3350 drives which were washing machine sized. The 3380 and 3390 were more refrigerator sized. There were two HDA's mounted edge side vertical in each cabinet. These cabinets formed a string of up to sixteen drives that was connected to a controller. Depending on the controller each string could be connected to two or four channels which improved throughput because the system could use whatever channel wasn't busy. I worked with 3390's when the "streaming I/O" channel was first introduced back in the late eighties. The standard channel had a capacity of 3MB/sec, the streaming I/O pushed that up to 6MB/sec. And it was all enterprise grade high reliability kit. Note to Dave: Among the reasons these stayed in service for so long was storage allocation. JCL (Job Control Language) was coded to allocate files with blocksizes that would efficiently fill a track on the drive. Move to a different device with a different track capacity and all of that JCL needed to be revised. Worse yet if you had a mix of devices in the shop you had to use a block size that worked well on each device. Allocations were typically done in tracks and cylinders (A column of tracks) so a cylinder on a 3380 was different from a cylinder on a 3390. If the allocation was too small the job could abort (abend) with an out of space condition (B37). When IBM introduced a drive array called a RAMAC (In honor of their original disk drive) in the late nineties they provided emulation of 3380 and 3390 devices so businesses wouldn't have to revise the JCL.
Awesome, thanks. Down the rabbit hole I go. So far found : en.wikipedia.org/wiki/History_of_IBM_magnetic_disk_drives and archive.computerhistory.org/resources/text/Oral_History/IBM_3380/IBM_3380.oral_history.2005.102657932.pdf
As a former IBM Field Engineer, I can tell you as you point out in the video overlay, that these drives were more in the 2-5Gb range and not 10 Meg. Also, the heads did not "ride" on the disk surface, but in fact were designed to "fly" on an air bearing that was molecules wide. The head NEVER touched the surface while spinning. The reliability of these drives was incredible and customers were glad (well glad might be a stretch) to pay even though they had to have enormous data centers to house the rows of hard drives. When you look at all the things that had to work correctly with little to no tolerance variation allowed... it is truly a marvel. And as you may have guessed, when you have the heads riding on an "air bearing" only molecules wide, the need to control expansion/contraction coefficients is very important.. hence the strip thermometer to assist with temperature control and problem diagnosis.
Listen to what he SAID. He SAID its a 10gig drive. HE had a 10meg or so drive in HIS computer around the same time this was made. Comparing and contrasting.
Actually folks I have been installing Halon 1301 systems since 1979. Contrary to popular belief, Halon 1301 does NOT suck or remove the oxygen concentration out of the room. Only when it hits a flame at about 900 F does it decompose into toxic gas. I have been in MANY discharge tests and stayed in the room for over 10 in. to run the gas analyser for the concentration test. It can give you a barn burner headache. But it does NOT kill you being in the room. Cardiac sensitization can occur in higher concentrations, or if you had a heart condition, you could be at a greater risk of danger, but not instant death. I think the movies exaggerate the myth. The detection system was cross zoned with ion and/or photo detectors that took 2 in alarm to start the 30 or 60 sec. delay. An abort button (dead man type) was easily accessible to halt the discharge before it released. But once it started to dump, no matter the amount of gas, it had to be all out in a nominal 10 sec. It is stored as a liquid, super pressurized to 360psi, and because of the low boiling point, vaporized at the nozzle instantly. It would fog a room, and visibility would be hampered for a few min. I always said the biggest hazard was not the gas discharge itself, but tripping and breaking your leg trying to escape. Safety was paramount, they would not let you dump a gas that removes (displaces actually) a gas into a room that could possible be occupied, just was not an option. CO2 will kill you quick, but is used in machinery spaces and the like, in non-occupied spaces only, never in data rooms. And of course, halon has been phased out many years ago, and I really don't think they are any systems left in service. Sorry, but I had to clear that myth up, it still blows my mind that so many people believe that, I have no idea how it started, no one ever has been killed in a halon dump, so I'm not sure why the myth lives on after decades. Anyway, thought some people might be interested in that myth. Awesome video as usual, love all your vids, have learned so much from you, keep it up, you da man!!
Actually there are some datacentre like facilities protected by CO2 though generally only ones housing extremely critical or very expensive (read supercomputers etc) equipment, Have only encountered like 3 of them during my career though and of those 2 were not even allowed to run in automatic mode period if personnel were present (Opening the door to the room disabled it until it was manually reset etc). The other was a very small room that was more than easily evacuated within the minute warning though even then there were cabinets containing O2 bottles with masks in the room but it was only to be used if getting out was for some reason not possible. Really no reason that should ever be the case though at most the door was like 15 meters (About 3 seconds at average walking speed) walk and the room is equipped with extremely hard to miss aural and visual alarms with a good 60 seconds warning time.
Halon works by displacing the atmosphere in the room (it's heavier than air), so yes, being in a room that is filled with Halon WILL suffocate you. You clearly weren't in a room that was filled with it. Yes, you can be in a room that is partially filled, but NOT in one that is filled with Halon above your head. The Halon systems on US Navy ships were designed to quickly and completely fill a space to put out a fire as quickly as possible. Not sure about the systems you worked with, but there is a reason that the "popular belief" exists.
Matt Elder Actually that’s not how halon works. CO2 systems displace oxygen. Halon interupts the ion exchange literally preventing the fuel from oxidizing. Unless a flame is present halon is far safer to humans, as you can walk around in it with nothing worse than a headache. CO2 on the other hand will kill you due to oxygen depletion. I have worked around many of both systems.
Matt Elder Well, keep in mind that the Navy may have been flooding the compartment with far more than a server room needs. The Navy has to consider that a room may have a gaping hole in conjunction with the fire.
I've got some ferrite core memory from an old mainframe used at a local nuclear research lab; it took 1 amp to switch 1 bit. Before they booted the mainframe, they had to call the local 2000MW power station (Didcot A) and tell them they'd be powering up in a couple of hours; can they please generate more? Insane.
I worked on an almost identical product at Control Data in about 1981-1982. I did very simplistic level failure analysis of "crashes" between the heads and platters.The leading edge of the head was beveled so that a layer of air would flow under the head. The head did not actually touch the platter when it was spinning, only when it was at rest. The space between the head and disk was just a few thousandths of an inch, if I remember correctly.The type I worked on sat on a cabinet about 4 feet high, 15 inches wide, and a couple feet deep. The cabinet held power supplies and a massive fan. The fan was not for cooling. The drive had big vacuum ports and the case would be vacuumed for many hours before being used, to remove dust and contaminants.These has a fairly significant failure rate in which the heads "stuck" to the platter when they were at rest. The surface of the head would "stick" and would damage the oxide when the spindle started to turn, ruining the platter and the information it contained.The units were assembled in a large clean room.The process of writing the "servo" tracks (the "locating" tracks) was quite tricky and sensitive. An oscilloscope was used by the technicians who calibrated the servo track writer weveral times per day. The servo writing machine was designed and built in-house at Control Data.I thought that I remembered that the capacity of the drive at the time was around 520 Megabytes, but I might be remembering that wrong. I just happened to have read a Wikipedia article about it a few weeks ago that indicated the Control Data product around the 1981 time period had a capacity of about 300 Megs. But that might have been increase in the first couple years after that.This was absolutely not a "plug and play" type machine. It was quite tricky to keep everything running properly during the manufacturing phase. And the fabrication and assembly of the drive in the clean room was very sophisticated.The discs had their final chemical coating (I don't know what it was) applied just hours before the whole assembly was put together. That chemical coating dip room was built adjacent to the the clean room. The discs were dripped, dried, then passed through the wall into the clean room.Control Data was a good company to work for in some ways, but in other ways it was horrible. My memory of those days is not pleasant. As Control Data started going down the tubes during the mid and late 1980's, the management got worse and worse and worse. The management were arrogant old stodgy paper pushers, who all got way too comfortable in their positions, and lost the energy of the earlier days of Control Data.When I start working there in about 1981, they had 55,000 employees. By the end, when they went out of business, they had about 300.There was a huge lawsuit between Control Data and IBM regarding intellectual property and patents, which I believe Control Data won. The suit lasted for years.
Really interesting stuff. I always love hearing about failure modes and the manufacturing hiccups for this kind of super expensive equipment, because it's not always obvious by first glance. Just because something has a giant price tag, that does not mean the end product is super reliable or that even manufacturing process was consistent.
" a layer of air would flow under the head. The head did not actually touch the platter when it was spinning" - I think that's true of modern drives also. The heads are said to fly above the disk surface. The distance isn't "a few thousandths of an inch" though - it's a good deal less than 0.001". Cigarette smoke particles are WAY too big to get underneath. The same applies to magnetic tape backup (I mean the type used in data centers), and tapes and their heads last orders of time longer if they are kept streaming, as opposed to start-stop operation, where the tape is continually landing on the heads. I know that engineering can produce absolutely incredible things for one-off projects - but in mass production, I honestly think engineering reached its peak of perfection in the modern hard disk drive - and also in the motor car. When you consider the complexity of a modern piston engine, it is incredible to me that I can get into my 11 year old car with the intention of driving from the UK to the South of Spain, with an expectation of getting there and back without mechanical failure. I have been using hard drives continually since 1985 as a field service engineer, and they have frozen in my car on Winter nights, and cooked in my car on Summer days, and been knocked about while carrying them, ever since the Compaq "sewing machine" portable PC, through the Grid, and on. I run them almost all day every day - and until now, I have never had a failure. I used to maintain RAID systems that had a couple of hundred drives in them, so I know that they do occasionally fail - but I've never had a personal drive fail.
@@DownhillAllTheWay If the head touches the disk - it would immediately scratch the hell out of it ferromagnetic layer that would ultimately destroy everything.
@@alexa.davronov1537 Indeed! I once went onto a customer site in Paris where one disk of a bank of 12, IBM 3330 disk drives had crashed. Look them up to see what they were like - but a multi-platter disk with 12 platters if memory serves. A disk drive stopped working, so the operator changed the disk - but both the disk and the heads in the drive were shot to pieces, so the new disk didn't work either, so he had seen two disks not working on the same drive - obviously a drive problem - so he put one of the disks onto a different drive - where it wouldn't work either (because the disk was crashed), and after that, he went into a mad disk-swapping frenzy, and by the time I arrived on site, he had crashed 11 of 12 drives. We bought some second-hand drives, because it was cheaper than getting replacement parts from IBM, and I used the heads from them - but I spent a loooong time setting them all up to align with the servo surface. Do you remember the "cat's eyes" on the oscilloscope? Ah, the good ol' days!
That vacuum was the reason the heads "stuck" to the platter when they were at rest. It's known problem in satellite design - metal parts in close contact could be "welded" one to another in vacuum. Sometimes you have to get parts moving just to prevent this.
In the early '70s I worked on Burroughs Disks. They were "Head per Track" Disks, no linear motor just one head over each track. Very fast. There were 4 disks with 8 surfaces. Each disk was about 4 feet in diameter and they would spin vertically. Each unit was 10MB. When we stated them up, we had to wait 24 hours for the speed to stabilise before we could write on them. They would run for about 10 years before the bearings would start to fail. Changing the bearings took about 3 days. Can't remember how much they cost but I am sure it was a lot.
Yes - the spindle was horizontal, so the disks spun in a vertical plane. While the bearings were being changed (or head assemblies), a "purge unit" was connected through two big flexible tubes to blow air that had been cleaned by passing it through "Absolute Filters". This way, all the time that the enclosure was open, there was a current of very clean air blowing through the unit, so that no dust could enter. One big problem they had was that the heads were mounted slightly away from the platters, and when the unit was running, the heads were pushed in by pneumatics, but the head assembly had "touch sensors", and if they ever detected a head touch, they would drop the pneumatics and retract the heads. We were getting so many touches that the disks were sent away to get a Teflon coating (for which the customer paid, of course). Then, the Teflon started flaking off and jamming under the heads, so they were sent away again for "decontamination" (for which the customer paid again). Later, I left Burroughs and worked for a third-party maintenance company, where I worked lots of stuff - including the IBM 3330 drives - very similar to the one in the video, but the heads recorded on both the top and bottom of each platter, and many were the hours I sat with an oscilloscope, setting up the cats eyes on these drives, to align the heads on each surface with the servo track on the bottom of a platter about half way up. They were good times - but I have to say - retirement is better!
Addendum: These big drives were used for the massive Mainframes of the day. I remember there were many field failure rates (of the drives). I remember when Tandem came out with individual mirrored drives (years later) it was considered brilliant revolutionary technology.I remember there were companies (customers) that sued Control Data over data loss on some of these big drives.
As someone else said they sounded like a jet. My uncle had a contract with the US airforce to dismantle and recycle these for disposal we used to shoot the platters out on the farm for target practice
that's what we called a Winchester Drive it was a sealed unit. I worked for DEC in NYC but DEC main office was in Maynard, MA USA. I worked for them in 1978-81 1982-1985 with a 18 month separation. I started as a Hard Drive specialist worked my way up to a Sr System engineer. I worked on several different HD's most were 20 14" platters (19 read heads 1 servo head) RP06, RP07 that were 300 MB. 95% of the DEC's HD's were made by CDC, the other 5% were made by DEC. I also worked on RK05 5 MB, RK06,7 15/30 MB, and a 90 MB Winchester/ removable HD it was a 30 MB removable & 60 MB Winchester HD. those were the days you used to troubleshoot to signal level or component level
Did you ever work on the RM05/CDC9766? I did a teardown of one (no platters), where the voice coil was massive. About the size of a soda can, and cylindrical (the coil alone). The magnet assembly was a cube the size of a stack of CD jewel cases. The die cast frame must've weighed almost a hundred pounds, if memory serves. As a 10yo I was in awe at the sheer amount of engineering that existed in one place. I remember applying 12v 1A to the voice coil, and it slammed so hard that my mom thought I ran into a wall!
IBM used the term Winchester to describe the 3340, not the 3390 (he is showing the HDA from a 3390). The 3340 had two spindles that each held one 30MB removable storage module. Two 30 MM Drives = 30-30 = Winchester 30-30, the popular lever action rifle that one the west. The 3390, 3380, 3370 and 3350 were not Winchester drives.
Funny fact, in Hungarian language when we speak we call all hard disk drives a "winchester" or "winyó" in general and we even write that in pretty much all non official conversations. Although the more official the situation gets the more likely you will write "merevlemez meghajtó" the literal translation and official word for "hard disk drive"... but even then there are only a few occasions when you would [and then you'd have to] say that (in some highly official situations like interrogation or court i can think of). The fact that computers were strictly restricted by the CoCom list until 1984 didn't help us to get people into computer science and even after '84 for years a Commodore or a ZX Spectrum was still a big thing for an average mid class family and totally out of reach for the majority of people. The fall of the iron curtain has changed things instantly and a thriving grey market of old replaced rigs mainly from Austrian companies made it somewhat more affordable. In 1990 I've got a non turbo 8088 512K RAM and a [compared to the one in the video) miniature 5.25" size Seagate ST225 20Mb "winyó" ... FFW 30 years and one photo I shoot with a pocket size camera exceed that capacity. :D
The server room I used to work in had a Halon flooding fire system. It was mildly terrifying every Monday morning when they tested the alarms in the building as you wandered if it wasn't by some chance the real thing and you were about to be suffocated.
Except that you wouldn't - It can be disorienting, and that's not a good thing in a fire, but Halon is not and, was never, toxic - that's a myth. It was phased out of use because of it's ozone depletion effect, not because of any safety concerns.
According to wikipedia it is toxic though reasonably mildly, it can 'affect the nervous system and other bodily systems'. Regardless of that it works by displacing oxygen and that means you won't be able to breathe, plus like you say the disorientaion plus the cold, the noise etc. I sure as hell wouldn't want to be in a room when it went off
The standard concentration for fire-fighting applications is in the 5-7% range - leaving plenty of oxygen for breathing - there is no danger of suffocation during a Halon discharge. If you read the Wikipedia article you will have noticed that it specifically states: Exposure to Halon 1301 in the 5 to 7 percent range produces little, if any, noticeable effect. The evacuation is standard fire-drill - there is presumably a fire so you need to go - NOW. There are no requirements to remove existing Halon fire-extinguisher system, and they remain in service in many places, but because Halon gas and Halon equipment is no longer commercially produced, once a system has been discharged, it will need to be migrated to alternative systems.
Sounds reasonable. I don't have the technical knowledge to argue otherwise and the ratio of oxygen only has to be disrupted to a limited amount to achieve a suppression of flames. My comment was only a throwaway one but I maintain I wouldn't want to be in the room when it went off, I'm certain it wouldn't be a pleasant experience. You're probably right that there is no requirement to replace unless it was discharged as thats consistent with many other decomissioning guidelines.
I had a friend in that used to work on these for jc-pennys. They kept all their prices on one of these... They called them Layer-Cakes. The would tear them apart and replace the platters... You would never replace a platter in a mode3rn drive...
EEVblog If you want to see some more sweet old school technology, look up "Inside Adam Savage's Cave: Geeking Out about Bits and Bytes" here on TH-cam. He has some interesting items in his collection.
My grandpa used to work on something very similar to these. If I recall correctly, he said they had 12 in. platters and stored 12 MB. I don't know if that was per platter or total. He said that they were designed to have the platters easily removed so that one drive could potentially use more than one set of platters. He also said the magnets on the linear actuators were so powerful that he ruined a very nice mechanical watch buy just putting his wrist near one.
Jeremy Catches You might if it breaks :) But thats quite unlikely and in this day and age someone would just buy a new one so you aren't completely wrong.
+seth094978 Holy shit. Now i'm wondering if the watch could have been degaussed. You know, with those big rings that they used on TV screens (CRTs) when they got magnetized by mistake. They started putting the rings in the sets to degauss them on the start up.
The 3390 Model 1 provided a storage capacity of 3.78 gigabytes to 11.35 gigabytes. The Model 2 had a capacity of 7.56 gigabytes to 22.7 gigabytes. The 3390 attached via the IBM 3990 Storage Control Models 2 and 3 to all ES/3090 models and IBM 308x models through an additional attachment feature. First customer deliveries of the 3390 Models 1 and 2 were scheduled for December 1989 at prices ranging from $90,000 to $275,000. A typical configuration consisting of three 3390 Model units was priced at $759,000.
I worked for IBM in the mid-90s in one of their outlets in Germany. The server room was the size of a tennis court, had a 3090 with IIRC 90 gig of HD space. The room was so noisy it was impossible to have a conversation, it was like stading next to a jet at takeoff. Unfortunately I don't remember the details of the mainframe -- I'm a UNIX guy responsible for the RS/6000 machines back then -- but I do remember that the whole floor was shaking from this beast.
by my math it's 8100b/cm², so about 90x90 dots in one square cm, it could be seen with an iPhone camera I bet (with the right lighting), and DEFINITELY a low-powered microscope
Hi Brings back memories from 1972 I used to work with large water cooled IBM main frames for a bank , 400hz PSU With loads of removable drives to enable data storage to bunker storage , we move 1tb of spinning storage up from London on weekend 8 Bullins air sprung truck to move it and 10 sparks to couple up the 60amp supplies to each bank we could have used less truck but the insurance value exceeded the value each one could carry . These had large three phase motors belt drive to vertical discs.. Total power for the floor was 1.5 megawatt I remember those units very well
Thats what we called a Winchesta, mate. in Goleta California they built them, employed alot of folks, Goleta had alot of hard drive industry making r/w heads too. I worked on the big Control Data removable pack drives from the 1970s, some 5 platter, some 10, big and the lid opens like a washing machine, put the plastic round thing with a handle down over the disk assembly, twist and lift it out and add the bottom to it so its sealed. had all the training, we would replace the heads and align them with an alignment pack, have to reach in there with a tool while spinning at 3000 rpm. I worked with 3.5 inch disk certifiers and would work on those in manufacturing and in clean rooms; also had a laser based disk profiler machine I helped develop, it could spin at 10,000 rpm on a glass air bearing spindle holding the disc with vacuum and have to reach in with alignment tools. Some customers would try and mess up and hit the disk? it would destroy the air bearing spindle that cost about $12,000. replacement was listed at 7-10 hours and took special training? I got where I could do the whole thing in under 20 minutes, a robotic assembly aligned to millionths of an inch that scanned top/bot of the disk, air/vacuum and all the electronics, I could replace the spindle and have it running in less than 20 minutes. But the Flash memory? pushed this technology out, company shut down
I look over at my PC with one of the fastest SSD's in the world inside (Samsung 840 Pro 256GB), but no matter how crazy fast it is, it will still never be as impressive as this hard drive.
One of these days a person will think about the 50,000 terabyte protein memory in their cerebral implant and say that it is not near as impressive as that old SSD from 2014.
There just is one little detail many forget: Soon, we won't need more storage for normal gamers/PC-users: People forget that the data has to be generated somehow and has to have a use. With 40GB/h of film, you have all the data you need - ever. as it would be enough to be at higher resolution than we can see, 3D, 100Hz and multi-language. Bigger films won't be any better - it would be far beyond human limits. Same for games: Once you are down to photo-realism, you won't need more storage for textures. Somebody gotta create the game-worlds too, and if we are at 200 different racetracks/Battlefields/Cities, you pretty much have a long time of playing ahead of you. And somebody gotta create those too. Maybe games will reach 200GB for a 40h game. But what after that? The sound will be better, than even the most trained person can distinguish, the films better than the best eye can notice, the games long enough to play a year. At that point, at that won't be that far away as it seems with all the tech-demos of photo-realistic engines, the vast majority won't need any bigger storage units.
***** As Re Said, we are already scratching at human limits. Take a 75" Screen. to NOTICE any difference, if you have perfect eyesight, you'd have to be at max 3m away. Likely, you won't sit that close to such a big screen. If you are any further away, it is virtually impossible to see any difference. Take 100" and you could go 4m away. Now your whole wall is one gigantic screen. And for a full movie in that resolution, you need about 80GB. Take it even further, and lets quadruple the resolution yet again, and we end up with film sizes of 320GB. I think, with a 200" monitor 3m in front of you, you got everything you ever need. And that is final. You now got full 3D, 100Hz, and perfect resolution, together with full surround, multilingual etc. After that, you only would notice if you got an even bigger screen and moved even closer. And that is something that 99.99% of all people will never do.
How is this video "painful" to watch? It is not a usable(or at least it does not make financial sense to use it) device today and it has little value. Instead of being tossed into a landfill, it served a final purpose by educating the viewer. "I died a little when you broke the seal and opened it." You must be new to the EEVblog.
Spocker spock True, but I'm not going to break the wheels off a Rolls Royce Silver Ghost either, even though it is clearly obsolete and has little (functional) value as a car today... I had to know these (logically) backwards, especially 3380's. Even though today's hardware is far superior, there are layers of abstraction that emulate these virtually still. So it would be a shame if I was asked what one was and could not find an example.
Previous to the voice-coil types, they used hydraulics to operate the head actuators. There were valves for distances of 1, 2, 4, 8, etc. Oil leaked over the drive deck and there was a little plastic bottle to catch the runoff. I took apart some of these old IBM voice-coil types, but with one big coil. The nut holding the platters together was TIGHT! The best part in it was the magnet. The magnet was the size of a building brick and was crazy strong. If you put two of them together, you needed a hammer and wooden blocks to drive them apart. If you put one behind the wall behind someone's CRT monitor, it would warp and distort the pix and drive them nuts. On H-P drives, they had a similar port, but it wasn't for Halon, it was for air. It would suck some air in through an "absolute" (meaning VERY fine) filter the size of a loaf of bread and blow a bit out the port. You had to measure the air pressure out every week, and when it started downhill, you knew to replace the filter, else you were cruising for a head crash. Halon just doesn't sound right for the IBM drive. There would be no ignition source in there. The platters for the bigger drives, say 14", made great backplates for nerdy wall clocks.
Absolutely not petabytes. I imagine Liam worked it out by multiplying the number of platters, and their size, and comparing that to modern hard drives. Since hard disk density is measured in bits / square centimetre. That and maybe he had a bit of a guess. What are they, 11 inch? Guess how many inches the spindle in the middle takes, and the unused area in the centre. Then do a bit of pi-r-squared, and compare that to modern hard drives, and multiply that by the highest modern drive's capacity. I dunno if it actually would be slow. Since, the further out in diameter you go, the faster the disk's linear speed is. Seek time might be a bit of a bugger, but that's what buffers are for. You could use an SSD for the buffer, or a few GB of RAM. Price wouldn't be a problem, if we're talking about a hypothetical 11 inch giganto-disk. As it is, usually HD storage on the petabyte scale uses arrays of 3.5" hard drives. In 1996 , Quantum released their "Bigfoot" series. They were 5 1/4" drives, years after everyone had switched to 3.5" drives. So they had much bigger platters, much higher capacity and a bit more speed. Since PCs had, and still have, 5 1/4" bays (for DVD drives these days, CD-ROM back in 1996), it wasn't a bad idea. In the end though the format disappeared after a while. I wonder why? Bigger diameters have several advantages, and every PC can fit them.
***** With technology you always have to decide when go "in". My first Intel-based PC in 1993 cost $3150 and had 33MHz 486, with 4MB of RAM and maybe a 80MB hard drive. But that was a white-box (not brand name) that suited my need to get into Windows and Linux software development. If I had waited say 2 years for something faster or cheaper, I would have lost all that time in developing skills that I used today. And as you rightly say, the gamble that manufacturers make to develop new tech and sell it at the leading edge to early adopters helps develop future products. And the customer is hopefully more competitive, for instance banks might have been able to offer loan approvals in hours instead of days which will make all the difference to their revenue.
Hi Dave, Some extra info. What you call the actuator is in fact the lock to lock the head carriage in the landing zone. The spring you mentioned is to bring the head carriage back in parking zone in case of a dramatic power failure The VCM is the thing that controls the location of the heads. One of the heads (the middle one of each device) is a so called servo-head. This head is the only head that knows where it is .;-) The surface of the servo plate is written in the factory and cannot be altered.(Servo head = Read only head) Very basically it work like this. Assume heads are on track 0 (Cylinder 0) Computer want to read data on track 110 of head 7. The disk will set a difference counter to 110. Then the VCM is told to move the heads in high speed towards track 110 Every time the servo head passes a servo track the difference counter is subtracted with 1. When the difference counter is reduced to a certain low number (say 20) the VCM wil start give several full reverse pulses to reduce the speed of the carriage. When difference counter is (e.g.) 10 the VCM will go in low speed for the last few tracks The moment the difference counter jumps to 0 the VCM will enter "Track following mode" That means it will lock itself (electronically) to this track. (We used to push the carriage manually to test the VCM poweramp and see if it could maintain track following) Head 7 will read the HA (Home Address) of the track and that should match the HA of Track 110 When HA is correct the data will be read. (there are some more records that will be read but that makes it to complicated for this post... and depends if you have a Count Key Data of Fixed block device) But basically it works like this. - B.t.w. Each carriage (VCM) is a standalone device. So this HDA has 2 devices in them.(in mainframes this is quite normal) In a mainframe of that time the device addresses works like this. The data path in mainframes goes like: CPU Channel Control unit Controller Device 1 number for Channel number 1 number for Control unit 1 number for device So a typical disk address would be 1AE. That would be channel 1 control unit A and device E Your HDA would then be addresses 1AE and 1AF - Mind you, this is how it worked in the 70's and 80's (time of this HDA)... now it is different.(Among things, larger (= more) addresses) - You are correct to assume that the datecode 1991 is because it was an replacement unit. Might even be that you still can order this device... Phone IBM and ask if they have a HDA with partnumber 26F7630 in stock. (Or ask a befriended IBM CE to check it... he has the correct phone number for this info) See FRU sticker at 3:54 (FRU stands for Field Replaceable Unit.) FRU numbers are the IBM technician part numbers to order a new part.
Well done mate. I wish there was some kind of filter I could apply to the comments on Dave's videos that only shows stuff like this. I'd love to see the controller for this, someone elsewhere in the comments suggested it was washing machine sized. The first HD I had was a 40MB Seagate 5.25" drive for which the controller was a full-length ISA card. I think the standard was called ST506. Also around 1991.
+Gavin Ward 3340 and 50 were indeed about washing machine size. 3330, 3380 and 3390 were much larger. (just look on internet for IBM 3330 DASD) 3330 wat the last DASD where the actual disk assembly (not heads!) could be removed by operator. - The controller for it just looks like the other disk but has controller logic in it beside the 2 disk addresses. Keep in mind that there are "Controllers" and "Control Units" Controllers serve actual disk clusters in the (Hex) range 0-F (It would serve for example 8 of these HDA's Dave showed us. Control Units serve Controllers and have address ranges like 00-FF (so controllers A0-AF, B0-BF, etc) Control units only talk to Controllers - And then there are Channels. Channels only talk to Control Units So the whole data path = CPU - Channel - Control Unit - Controller - Device (one of the 2 moving head carriages Dave showed. Each HDA has actually 2 disks in them!))
Hi Spikejwh1 great 👍 Could well be to order those drives. I remember that we even had to rebuild scrapped production equipment to produce some more IBM62PC piccolo drives for special customers. Obviously it didn't matter how much it costs.
I remember those. I used to work on IBM Midrange systems (the System/36 range of computers). the hard drive platters were 3 feet across and weighed 40 kilos. we usually had one die every 2 years or so. 350MB of storage on the model B and 768MB on the Model D. the actual machine was bigger than a washing machine and made a horrible din as fans and motors ran in it 24 hrs a day. My office had 2 of them and every summer, the air chiller unit would fail and tempretures would shoot up to over 100 degrees. those were the days when computer techs were working at the coal face!
Worked in an IBM shop during this period. He's handling a 3390 drive unit. They almost had the same capacity of the 3380s that they replaced, 630Meg, 1260M, or 1890M per actuator/address, for the standard, "D", or "E" models. The 3390s saved 'footprint' space in the data center for the same storage, along with power and noise. A 3380 drive had a sticker on the side, > 70 kgs, recommending 2 CEs to swap them out. An actuator on each side gave the device 2 addresses. Two drives per box, total of 8 boxes per string. The 3390 has two actuators / drive also, as seen in the video. Top & Bottom. If it failed, a single CE could unplug it, slide it out, and replace with a new one. "CE" - IBM Customer Engineer.
Pshaw. That's nothing. An ex-boss of mine used to work at Control Data when they were developing a no-holds-barred disk drive. One platter, six feet across, with 8,192 fixed heads. Cost of around $18 million 1978 dollars. Not a problem for their Rodeo-drive customers like the NSA, Westinghouse and Livermore. They were clamoring for disks with low latency, so no head-stepping. One day he came in to work about midnight, typical hardware hacker time, and the building was almost impossible to enter due to a loud screeching wail sound. The big disk had a head crash, which cascaded from head to head, trashing all the heads and the disk surface. They never had the nerve to try building a second one. It had lasted all of three days and nights.
1/4 mil doesn't even come close. These units were used in RAID arrays that had many of these units. What you have was THE field replaceable unit (FRU). Repairs would have been made in a clean lab, probably in Endicott. The heads actually flew by bernoulli effect over the platters and head to disk interference (HDI) was often a terminal event, nevertheless, stuff did happen and you could swap a unit in the raid and the raid would rebuild the data from the other drives. Then send the offending unit back for rebuild. There were also some aeronautical/defense applications in which units similar to these were used. I could tell you more, but I can't.
So many errors and misunderstandings in this. For starters:- That's just the HDA (Head Disk Assembly) from an IBM 3390, a box the size of a couple of fridges. There were 2 to 6 HDAs (depending on model) in the box, which also contained all the other electronics required to provide the equivalent function to the familiar hard disk. These were introduced in 1989, and the HDAs have absolutely no resemblance whatsoever to the earlier models that they replaced. (I've torn down a number of them myself) This is definitely NOT 1970's technology.. And the HDA cost nothing like $250,000. The whole box, along with the other HDA(s) and electronics, started at $90,000.
I was irritated at the '4Tb in 1.8" = the smallest form factor' statement at first. I was like "Don't we have M.2-SSD's with much more capacity for a bunch of years now?". Then I looked at the upload date... 😅
This thing in today's market would cost $848,398 AUD, or around $636,298.50 USD. So basically something JayzTwoCents would put in Little Jay's computer because she wants to get into MSPaint.
Or around $40 as a non functioning decoration since that is all it's good for. At 5MB I don't think that it would've been able to hold MS Paint from Windows 95 but is easily large enough to store all of Jay Z's good songs. Physically, the case is almost large enough to store all his celebrity endorsements
For me its kind of funny seeing that "old iorn" again. My first day on the job in field service was in 1976 . We were installing "field change orders" on 10 megabyte ,full size disk drives (about the size of a washing machine) The disk drives had a linear bobbin type motor. The field change was adding an RC circuit to the linear motor circuit so that the disk drive would not pickup and play music from the local am radio stations. ...yep ,in some installations it was an issue lol.
I worked on these things and the filters actually filtered recirculated air within the unit, not outside air. Additionally, these were were used with mainframe computers like the IBM 360/370 series. Servers as we have today, didn’t exist yet. The platters were double sided except for the top and bottom platter.
You need to remember that before IDE harddrives, drives were dumb and had their circuitary as part of the controller.The I in IDE stood for Integrated.
I thought only the vacuum tube era was that massive. But this is the transistor era already. Next time I play Quake 1 or watch Alien I will not complain about the computers everywhere wondering what the heck they are for and why they sounds like a factory...
OK 10MB is really nothing nowadays. But if you make one of these today (same shape and size) with the capacity of 10PB (Petabytes) with a sticker price of $50,000 USD, people would start drooling all over it like they did in the 70s.
that sensor in the actuator is the "Home" sensor, lets the drive know when the heads are over the "Zero" track.. so the drive knows where to start counting from ..excellent video, haven't seen that type of drive in over 20 years.. please keep these types of video's coming.. btw those heads have more than 1 gap, there is an erase head there is a r/w head, and there is also a dual head that is called a guard head which makes a tunnel on either side of the data channel so adjacent channels do not interfere with each other..
I find it a little sad that now we're all moving to SSD, I mean, possibly the finest piece of mechanical-electric technology you'll ever own, compared to a few chips on a PCB.
6:53 "Otherwise, the halon gas would kill you." I thought this too, then I searched a bit and found this isn't exactly the case. Yes, gases can be dangerous, but not all of them kill datacenter workers.
@@pedroalexander8921 Yes It displaces the air, which is how the fire suppression work. also because of the energy lost to convert the cold liquid gas, it would chill a whole room with the phase change from liquid to gas. just like a CO2 fire extinguisher, the temp drop forms snow ice and dry ice, it's just under enough pressure to keep it liquid, and it occupies a small space so it transports easier.
IBM generally had a controller with the smarts connected to a bunch of not so intelligent drives so I would not expect this item to have much smarts. My first hard drive was from IBM and had 4 Megabytes on a stack of platters wider than this one. Amazing stuff. THESE are the good old days for computers and other hardware. Also, the top head was frequently a sync track. Lots of little markers for timing, track start, sector start, etc. Not a read/write surface.
Samsung and SanDisk have been selling 1TB Micro SD cards for over a year now, so that’s 1,000,000 MB, that’s 100,000 times more memory in something that weighs roughly 82,000 times less. Yes, that’s right, this 10MB hard drive weighs roughly the same weight of 82,000 Micro SD cards, or equivalent to a measly 82 Petabytes!
The earliest IBM commercial hard drive had something like 50 surfaces (about 26 platters, further separated than the ones in this unit) and ONE head assembly which moved in and out and, while completely out of the platter area, up and down. Next they came up with the multiple head access “comb,” and made the storage media removable, allowing separate data storage for multiple systems (or clients). The most popular ones had 10 or 20 surfaces, and were stored offline in two-part carriers with a handle on top. But smaller and smaller head gaps made the removable (by the operator between jobs) system too unreliable, especially when people smoked in the computer rooms (an ash particle hit by a flying head was like a rock hit by a lawnmower), so these “head drive assemblies,” or HDAs, were the solution. Each HDA was a separate volume, and by that time, high speed tapes were fast and cheap enough to “swap drives” by copying the current content to a backup tape, then loading another backup tape over it. It took more time than manually swapping removable packs, but not much more, since there was no need to spin the drive down and back up. These modules were replaceable only by the CE (IBM speak for repair tech) and required the entire controller and all volumes to be powered down and the cover removed. And notice how the access arms move in and out along a radius of the spindle? Not so good for PC sized drivers, because of the noise and vibration (sometimes the big old drives shook like unbalanced top loading washers and driers), and the extra space needed for the fully withdrawn access arms. Today’s access arms swing around a shaft in the corner of the rectangular housing like a phonograph tone arm; the deviation from a straight radial line isn’t large enough to make a difference.
The 64th Shadow Sry I was in a hurry in real we won't have a smaller structure than 1- 0.5 nm because for transistores they make no sense.. The biggest harddrive at the moment is 6TB, so maths say this 14 nm*(1/0.5 nm) * 6TB = roughly 168 TB is the biggest possible HD with in the normal specification of a normal HD in these days
Fiercesoulking you forgot to calculate in 2 dimensions. Shrinking from 14nm to 7nm brings you 4 times as many bits. Shrinking from 14 to 1 brings you 288 times as many bits. And there is still the possibility to create thinner discs and heads to fit more discs into one harddrive case. Even without smaller structures on the disc itself this may also increase capacity. One could also try to store more bits at a single location so that a location is not 0 or 1 but rather 0, 1, 2 or 3 (0, 33, 66 or 100% magnetic force; hdds already use this technique but I'm sure there is still some potential). So we do a simple calculation: Half the structure size (7 instead of 14nm), twice as many discs and twice as many bits per location would raise the capacity from 6 TB to 96 TB (6TB x 2 x 2 x 2 x 2)
That motor you removed from the bottom, before you could pull out the ACTUATOR assembly is the head LOAD/UNLOAD motor. It was used to remove the heads from the disk before the disks stopped at power down. If the motor was allowed to come to a stop with the heads in contact with the disk they would STICK. The LOAD/UNLOAD motor also LOCKS the actuator in the PARK position when the motor is stopped.
I’ve taken apart and repaired one of these. They’re sooooo heavy. We still have the drive platters on our office entrance. I got mine as part of an old system 360. Your drive is a bit newer than the one I had, but it was... 30 years ago. So 🤷♂️.
Most people do not realise how close the heads floated above the disk surface. If the head was scaled up to be the size of a 747 and the air gap was scaled up accordingly, the 747 would fly into strawberry.
First question is why so many downvotes second question I'd love to see this thing and action it's a beautiful piece of equipment I hope that the disassembly did not destroy it I know that you have to be extra careful with hard drives although I'm no expert
ojkolsrud1 I wonder if it is safe to say a terabyte of RAM can fit in there? I also wonder how much storage you can fit, along with the amount of platters you can fit. It's amazing how technology has changed!
KrazyLemming and still has to. If you look at biology 1TB isn't that much. The human genome, as far as I know, consists of about 3GB of data. Now that is even much when comparing it to flash memory! And when thinking about what one ejaculation consists of: over 100TB of data! WOW! Incredible. I don't know whether we will ever reach that high data density, but well, if we do, then we appear to have something worth waiting for :)
Maximilian Mustermann 100TB in an ejaculation is a number that often circulates on the internet, but it seems to me that it's only half a human genome (1,5-2GB) of data, just with crazy redundancy..
Bart Guliker Well, if you check the number s: 3.2 billion bases (I don't know the exact english term). The 100-200TB Idea isn't all off. I was far from exact! :)
It's not certain that it's connection port is even IDE, you know. In that case, what you propose would be difficult! But, he should at least tried hooking it up to some power so we could watch it spin up!
Its not possible to do so. He would need a System/390 mainframe as this is an IBM drive and would only work with IBM mainframes, the drive controller (this is not an IDE drive so has no onboad controller), a S/390 OS and software capable of running on the mainframe to run the controller and understand the filesystem and file formats. Basically if he had an S/390 mainframe to put this in it would take up his entire workshop and powering it up would require work too.
XmarkedSpot No, microdrives were actual hard disk drives with a 1 inch platter. Same case as a CF card (thick version), but back then when they were a thing they had a much larger capacity than one could have at a reasonable price on Flash-based CFs.
XmarkedSpot I bought one a good 10 years ago. Was fine even if rather slow, and it did what hard drives like to do best and failed on me after only a few months. Can't remember what I was using it in though. At least I got to take it apart and marvel for a while at the impressively tiny machinery... quite the opposite of the one in the video, lol.
kilrahvp I got mine with yet unseen 2GB for my first-gen digital camera (~2003 Canon, capturing a whopping 3.1 megapixels). The drive actually outlived the camera (8yrs of heavy use) and who knows, probably still might work. I guess it would have failed me if i wouldn't have a backup routine ;) Having a 64GB sdxc by now doesn't really make me miss it... but i agree, the technology still remains very impressive.
For the record 250,000 USD at the time of the drive's production is around 800,000 current USD. If inflation wasn't already calculated, then this is a better number for scale perception. In other words, that's really expensive!!!
At the cost rate of that unit, a 1 gigabyte hard drive array in that day, would have cost: 25 Billion Dollars! At the cost rate of that unit, a 1 terabyte hard drive array in that day, would have cost : 2.500000000×10¹³ Dollars! W O W !
This hard drive was $250,000 and could store 1.89/3.78 GB, so a 1GB drive would have costed anything between $66,000 and $132,000. So a 1TB drive would have costed anything between $66,137,500 and $132,275,000.
Wow, takes me back. Used to work on GE Medical CAT scanners (circa 1981) that stored scan data on CDC 96MB removable platters, the entire drive enclosure was about the size of a small washing machine. Replacing a read/write head in the field was a treat; no clean room, basic hand tools and an old school Tektronix storage oscilloscope. Was only required to do it twice but I can say the installation and calibration instructions provided by CDC were outstanding (provided on Microfiche slides)!
I have an unexplainable but very real soft spot for old hard-drives the way some people do for puppies and things. I just love them! Plus those mid 1990s hdds have the best magnets to re-purpose to keep even he heaviest grocery list stuck to the fridge. Hehe.
I worked as a student employee at IBM in Rochester, Minnesota, USA between Oct. 1989 and Aug. 1990. I worked in the platter manufacturing area. The section I was in produced 3.5 inch and 5.25 inch platters.One of the techs on in the coating area told me that one of the 5.25 platters would hold up 125 Meg of data. I was amazed because at that time I wanted to get an IBM desktop which the largest hard drive available was around 60 Meg. Another department produced the 11 inch and 13 inch platters. You estimate how much the larger platters would hold based off of what I have described. I was very cool working there because I got to see how they were made from blanks all the way through to final test. The final test area had about 10 quad platter testers. In that area we would have to change out the heads on the testers. We would inspect them under a microscope and clean them. A friend of mine worked in another area when they made the heads. They would manufacture a long block and laser cut the heads from that. Very cool experience!
BTW the thermometer strip is a standard-looking LCD thermometer. The numbers are printed underneath, and each segment of the LCD becomes transparent / opaque at particular temperatures, that's how it works. For the heads, I think the reason there's 4 contacts, with only 2 being used, is so they can use the same heads upside-down, which would reverse the L / R direction. Saves them having to make and use 2 types otherwise. If that's just 10MB, then the data density on the discs is going to be pretty close to that of early '80s floppy drives. Which wasn't much!
I think pc hardware barely evolved in the last 7 years...cpu are only 20-25% more powerfull,4-8gb of ram is still standard,ddr3 is still here,hard drive capacity went from 320-500gb to 500gb-1tb,we still use Sata and besides graphic cards getting a huge boost,nothing else dramatically changed...
That's exactly what I was thinking. Even two years after your comment, almost nothing changed. Sure, they got a bit faster and stuff, but nothing exciting, just CPU bugs :D
Well, if you have a lot of money you can try with msata 1Tb SSDs, they are 51 x 30 x 4mm big so multiply that by 1000, you get a petabyte of crazy fast storage for about 250 000€ in a 32*60*32cm box Yeah, pretty sick tech that we have today, you could probably fit 1 of these petabyte boxes in that bulky IBM HDD enclosure. Also, considering that the internet 1.2 Zetabytes big, you would need 1.2 Million of these boxes units. It would be about 0,06144 cubic meters per enclosure, multiply that by 1 200 000 that makes 73 728 cubic meters, A 747 has a cargo + passenger volume of about 1000 cubic meters so that's 73 Boeing 747 to transport the entire internet And AirFrance has exactly 73 of these jumbo jets... It would cost 1 200 000*250 000 = 300 B€ of SSDs and it would weight about 10 200 tons of SSDs, and Shell can have this amount of money... After all of this math, the conclusion is: AirFrance and Shell are going to fuse together to have enough money to buy all of the SSD's they need to put them in the AirFrance's jumbo jet fleet, send them to Australia and then they will use EMP bombs to destroy the internet and keep their copy of the internet in hostage to get anything they want in return. My brain hurts so bad
I managed a datacenter for 15 years and the halon was no joke, we had training and were told if the alarms went off to vacate the room asap, injury could occur as it "replaces" the oxygen in the room.
Looks like an IBM 3970, and there would be a room full of these for banks, etc., you provided an excellent But they were built for super reliability, and IBM built them like tanks. I used to program mainframes and super-computers, and this was certainly a pleasant walk down memory lane! Thanks again !!
When he said, "Hi - welcome to teardown...Tuesday", it reminded me of PeeWee Herman. And how 'bout a testosterone shot to get the pitch of your voice down into the "man" range...
Actually he is a shoo-in for Dick Simnel if Raising Steam ever gets the Tom Hanks/Spielberg treatment. All that and a top tenor and you want it treated with chemicals? Silly man.
I was going to type the same thing but I honestly didn't want to be mean or seem like a troll! Nothing a little testosterone couldn't fix, that or voice training.
![ร้อยพ่อพันแม่ - WanMai [Official MV]](http://i.ytimg.com/vi/XY2pNYASEbg/mqdefault.jpg)
Ha, ha! I worked on 14-inch Winchester drives in the early 1980s. The unit shown in this video isn't the complete drive; it's just the head-disk assembly (HDA). The full drive, with power supplies and electronics cards was about the size of a washing machine, and that still wasn't enough to get the data on and off the discs, as the drives had to be connected to a controller unit that handled a string of 8-16 drives. This drive is more advanced, having an integral spindle motor, but the 14-inch drives that preceded it in the 1980s were belt-driven by 3-phase induction motors about the size of what you would find in a washing machine. The systems needed an air-conditioned room in order to work reliably. It's amazing that today's hard drives have all of that crammed into a pocket-size package with orders of magnitude higher reliability and lower power consumption!
+ebulating I find a 300GB microSD more interesting.
Thank you so much for commenting. I knew I wasn't crazy. In 1998 I had to go on site to some railyard at night for a quote-unquote hard drive replacement. Arriving in a room with no lights I have to use my flashlight. I open a crate and I swear among all the padding I saw a giant belt-driven something that looks like a sewing machine. I slowly closed the box and backed away.
Pretty weird when you're expecting to find something the size of a VHS tape or smaller. I actually found this video looking for what that was. I'm looking up Legacy hard drives Etc. If you have any Google image links or machine type descriptions for what you think that was I saw or something related, please reply. Again thanks.
You're thinking of the 3330 and 3350 drives which were washing machine sized. The 3380 and 3390 were more refrigerator sized. There were two HDA's mounted edge side vertical in each cabinet. These cabinets formed a string of up to sixteen drives that was connected to a controller. Depending on the controller each string could be connected to two or four channels which improved throughput because the system could use whatever channel wasn't busy. I worked with 3390's when the "streaming I/O" channel was first introduced back in the late eighties. The standard channel had a capacity of 3MB/sec, the streaming I/O pushed that up to 6MB/sec. And it was all enterprise grade high reliability kit.
Note to Dave: Among the reasons these stayed in service for so long was storage allocation. JCL (Job Control Language) was coded to allocate files with blocksizes that would efficiently fill a track on the drive. Move to a different device with a different track capacity and all of that JCL needed to be revised. Worse yet if you had a mix of devices in the shop you had to use a block size that worked well on each device. Allocations were typically done in tracks and cylinders (A column of tracks) so a cylinder on a 3380 was different from a cylinder on a 3390. If the allocation was too small the job could abort (abend) with an out of space condition (B37). When IBM introduced a drive array called a RAMAC (In honor of their original disk drive) in the late nineties they provided emulation of 3380 and 3390 devices so businesses wouldn't have to revise the JCL.
Give it a few years and this stuff will be crammed into something the size of your thumbnail
Awesome, thanks. Down the rabbit hole I go.
So far found :
en.wikipedia.org/wiki/History_of_IBM_magnetic_disk_drives
and
archive.computerhistory.org/resources/text/Oral_History/IBM_3380/IBM_3380.oral_history.2005.102657932.pdf
As a former IBM Field Engineer, I can tell you as you point out in the video overlay, that these drives were more in the 2-5Gb range and not 10 Meg. Also, the heads did not "ride" on the disk surface, but in fact were designed to "fly" on an air bearing that was molecules wide. The head NEVER touched the surface while spinning. The reliability of these drives was incredible and customers were glad (well glad might be a stretch) to pay even though they had to have enormous data centers to house the rows of hard drives. When you look at all the things that had to work correctly with little to no tolerance variation allowed... it is truly a marvel. And as you may have guessed, when you have the heads riding on an "air bearing" only molecules wide, the need to control expansion/contraction coefficients is very important.. hence the strip thermometer to assist with temperature control and problem diagnosis.
Listen to what he SAID.
He SAID its a 10gig drive.
HE had a 10meg or so drive in HIS computer around the same time this was made.
Comparing and contrasting.
Damn
@@sugarnads 2:06 he says 10 MB. @Emtdrummer was correct.
We're these the diamond thin film heads back then?
Actually folks I have been installing Halon 1301 systems since 1979. Contrary to popular belief, Halon 1301 does NOT suck or remove the oxygen concentration out of the room. Only when it hits a flame at about 900 F does it decompose into toxic gas. I have been in MANY discharge tests and stayed in the room for over 10 in. to run the gas analyser for the concentration test. It can give you a barn burner headache. But it does NOT kill you being in the room. Cardiac sensitization can occur in higher concentrations, or if you had a heart condition, you could be at a greater risk of danger, but not instant death. I think the movies exaggerate the myth. The detection system was cross zoned with ion and/or photo detectors that took 2 in alarm to start the 30 or 60 sec. delay. An abort button (dead man type) was easily accessible to halt the discharge before it released. But once it started to dump, no matter the amount of gas, it had to be all out in a nominal 10 sec. It is stored as a liquid, super pressurized to 360psi, and because of the low boiling point, vaporized at the nozzle instantly. It would fog a room, and visibility would be hampered for a few min. I always said the biggest hazard was not the gas discharge itself, but tripping and breaking your leg trying to escape. Safety was paramount, they would not let you dump a gas that removes (displaces actually) a gas into a room that could possible be occupied, just was not an option. CO2 will kill you quick, but is used in machinery spaces and the like, in non-occupied spaces only, never in data rooms. And of course, halon has been phased out many years ago, and I really don't think they are any systems left in service. Sorry, but I had to clear that myth up, it still blows my mind that so many people believe that, I have no idea how it started, no one ever has been killed in a halon dump, so I'm not sure why the myth lives on after decades. Anyway, thought some people might be interested in that myth. Awesome video as usual, love all your vids, have learned so much from you, keep it up, you da man!!
Actually there are some datacentre like facilities protected by CO2 though generally only ones housing extremely critical or very expensive (read supercomputers etc) equipment, Have only encountered like 3 of them during my career though and of those 2 were not even allowed to run in automatic mode period if personnel were present (Opening the door to the room disabled it until it was manually reset etc). The other was a very small room that was more than easily evacuated within the minute warning though even then there were cabinets containing O2 bottles with masks in the room but it was only to be used if getting out was for some reason not possible. Really no reason that should ever be the case though at most the door was like 15 meters (About 3 seconds at average walking speed) walk and the room is equipped with extremely hard to miss aural and visual alarms with a good 60 seconds warning time.
Halon works by displacing the atmosphere in the room (it's heavier than air), so yes, being in a room that is filled with Halon WILL suffocate you. You clearly weren't in a room that was filled with it. Yes, you can be in a room that is partially filled, but NOT in one that is filled with Halon above your head. The Halon systems on US Navy ships were designed to quickly and completely fill a space to put out a fire as quickly as possible. Not sure about the systems you worked with, but there is a reason that the "popular belief" exists.
Matt Elder Actually that’s not how halon works. CO2 systems displace oxygen. Halon interupts the ion exchange literally preventing the fuel from oxidizing. Unless a flame is present halon is far safer to humans, as you can walk around in it with nothing worse than a headache. CO2 on the other hand will kill you due to oxygen depletion. I have worked around many of both systems.
@@ericapelz260 that wasn't how it was explained to us when I was in the Navy 30 years ago, but I'll admit that my memory could be faulty.
Matt Elder Well, keep in mind that the Navy may have been flooding the compartment with far more than a server room needs. The Navy has to consider that a room may have a gaping hole in conjunction with the fire.
I've got some ferrite core memory from an old mainframe used at a local nuclear research lab; it took 1 amp to switch 1 bit. Before they booted the mainframe, they had to call the local 2000MW power station (Didcot A) and tell them they'd be powering up in a couple of hours; can they please generate more? Insane.
Should just made their own reactor for it if they were a nuclear lab ahahaha
I worked on an almost identical product at Control Data in about 1981-1982. I did very simplistic level failure analysis of "crashes" between the heads and platters.The leading edge of the head was beveled so that a layer of air would flow under the head. The head did not actually touch the platter when it was spinning, only when it was at rest. The space between the head and disk was just a few thousandths of an inch, if I remember correctly.The type I worked on sat on a cabinet about 4 feet high, 15 inches wide, and a couple feet deep. The cabinet held power supplies and a massive fan. The fan was not for cooling. The drive had big vacuum ports and the case would be vacuumed for many hours before being used, to remove dust and contaminants.These has a fairly significant failure rate in which the heads "stuck" to the platter when they were at rest. The surface of the head would "stick" and would damage the oxide when the spindle started to turn, ruining the platter and the information it contained.The units were assembled in a large clean room.The process of writing the "servo" tracks (the "locating" tracks) was quite tricky and sensitive. An oscilloscope was used by the technicians who calibrated the servo track writer weveral times per day. The servo writing machine was designed and built in-house at Control Data.I thought that I remembered that the capacity of the drive at the time was around 520 Megabytes, but I might be remembering that wrong. I just happened to have read a Wikipedia article about it a few weeks ago that indicated the Control Data product around the 1981 time period had a capacity of about 300 Megs. But that might have been increase in the first couple years after that.This was absolutely not a "plug and play" type machine. It was quite tricky to keep everything running properly during the manufacturing phase. And the fabrication and assembly of the drive in the clean room was very sophisticated.The discs had their final chemical coating (I don't know what it was) applied just hours before the whole assembly was put together. That chemical coating dip room was built adjacent to the the clean room. The discs were dripped, dried, then passed through the wall into the clean room.Control Data was a good company to work for in some ways, but in other ways it was horrible. My memory of those days is not pleasant. As Control Data started going down the tubes during the mid and late 1980's, the management got worse and worse and worse. The management were arrogant old stodgy paper pushers, who all got way too comfortable in their positions, and lost the energy of the earlier days of Control Data.When I start working there in about 1981, they had 55,000 employees. By the end, when they went out of business, they had about 300.There was a huge lawsuit between Control Data and IBM regarding intellectual property and patents, which I believe Control Data won. The suit lasted for years.
Really interesting stuff. I always love hearing about failure modes and the manufacturing hiccups for this kind of super expensive equipment, because it's not always obvious by first glance. Just because something has a giant price tag, that does not mean the end product is super reliable or that even manufacturing process was consistent.
" a layer of air would flow under the head. The head did not actually touch the platter when it was spinning" - I think that's true of modern drives also. The heads are said to fly above the disk surface. The distance isn't "a few thousandths of an inch" though - it's a good deal less than 0.001". Cigarette smoke particles are WAY too big to get underneath. The same applies to magnetic tape backup (I mean the type used in data centers), and tapes and their heads last orders of time longer if they are kept streaming, as opposed to start-stop operation, where the tape is continually landing on the heads.
I know that engineering can produce absolutely incredible things for one-off projects - but in mass production, I honestly think engineering reached its peak of perfection in the modern hard disk drive - and also in the motor car. When you consider the complexity of a modern piston engine, it is incredible to me that I can get into my 11 year old car with the intention of driving from the UK to the South of Spain, with an expectation of getting there and back without mechanical failure.
I have been using hard drives continually since 1985 as a field service engineer, and they have frozen in my car on Winter nights, and cooked in my car on Summer days, and been knocked about while carrying them, ever since the Compaq "sewing machine" portable PC, through the Grid, and on. I run them almost all day every day - and until now, I have never had a failure. I used to maintain RAID systems that had a couple of hundred drives in them, so I know that they do occasionally fail - but I've never had a personal drive fail.
@@DownhillAllTheWay If the head touches the disk - it would immediately scratch the hell out of it ferromagnetic layer that would ultimately destroy everything.
@@alexa.davronov1537 Indeed! I once went onto a customer site in Paris where one disk of a bank of 12, IBM 3330 disk drives had crashed. Look them up to see what they were like - but a multi-platter disk with 12 platters if memory serves. A disk drive stopped working, so the operator changed the disk - but both the disk and the heads in the drive were shot to pieces, so the new disk didn't work either, so he had seen two disks not working on the same drive - obviously a drive problem - so he put one of the disks onto a different drive - where it wouldn't work either (because the disk was crashed), and after that, he went into a mad disk-swapping frenzy, and by the time I arrived on site, he had crashed 11 of 12 drives. We bought some second-hand drives, because it was cheaper than getting replacement parts from IBM, and I used the heads from them - but I spent a loooong time setting them all up to align with the servo surface.
Do you remember the "cat's eyes" on the oscilloscope? Ah, the good ol' days!
That vacuum was the reason the heads "stuck" to the platter when they were at rest. It's known problem in satellite design - metal parts in close contact could be "welded" one to another in vacuum. Sometimes you have to get parts moving just to prevent this.
In the early '70s I worked on Burroughs Disks. They were "Head per Track" Disks, no linear motor just one head over each track. Very fast. There were 4 disks with 8 surfaces. Each disk was about 4 feet in diameter and they would spin vertically. Each unit was 10MB. When we stated them up, we had to wait 24 hours for the speed to stabilise before we could write on them. They would run for about 10 years before the bearings would start to fail. Changing the bearings took about 3 days. Can't remember how much they cost but I am sure it was a lot.
""Head per Track" Disks" YEEAH! None of that linear motor rubbish!
@@johanveltheim4331 hahahahahaha
Incredible.
Yes - the spindle was horizontal, so the disks spun in a vertical plane. While the bearings were being changed (or head assemblies), a "purge unit" was connected through two big flexible tubes to blow air that had been cleaned by passing it through "Absolute Filters". This way, all the time that the enclosure was open, there was a current of very clean air blowing through the unit, so that no dust could enter.
One big problem they had was that the heads were mounted slightly away from the platters, and when the unit was running, the heads were pushed in by pneumatics, but the head assembly had "touch sensors", and if they ever detected a head touch, they would drop the pneumatics and retract the heads. We were getting so many touches that the disks were sent away to get a Teflon coating (for which the customer paid, of course). Then, the Teflon started flaking off and jamming under the heads, so they were sent away again for "decontamination" (for which the customer paid again).
Later, I left Burroughs and worked for a third-party maintenance company, where I worked lots of stuff - including the IBM 3330 drives - very similar to the one in the video, but the heads recorded on both the top and bottom of each platter, and many were the hours I sat with an oscilloscope, setting up the cats eyes on these drives, to align the heads on each surface with the servo track on the bottom of a platter about half way up.
They were good times - but I have to say - retirement is better!
To think that modern HDDs need less than 4s to spin up and stabilize, amazing how far and fast computer tech has progressed.
Addendum: These big drives were used for the massive Mainframes of the day. I remember there were many field failure rates (of the drives). I remember when Tandem came out with individual mirrored drives (years later) it was considered brilliant revolutionary technology.I remember there were companies (customers) that sued Control Data over data loss on some of these big drives.
Would have been fun to hear that spin up
They sounded like a jet taking off. And they spun up one at a time, so got to hear it multiple times in a large installation.
They actually generated quite a bit of bearing, platter, and head noise.
As someone else said they sounded like a jet. My uncle had a contract with the US airforce to dismantle and recycle these for disposal we used to shoot the platters out on the farm for target practice
that's what we called a Winchester Drive it was a sealed unit. I worked for DEC in NYC but DEC main office was in Maynard, MA USA. I worked for them in 1978-81 1982-1985 with a 18 month separation. I started as a Hard Drive specialist worked my way up to a Sr System engineer. I worked on several different HD's most were 20 14" platters (19 read heads 1 servo head) RP06, RP07 that were 300 MB. 95% of the DEC's HD's were made by CDC, the other 5% were made by DEC. I also worked on RK05 5 MB, RK06,7 15/30 MB, and a 90 MB Winchester/ removable HD it was a 30 MB removable & 60 MB Winchester HD. those were the days you used to troubleshoot to signal level or component level
Did you ever work on the RM05/CDC9766? I did a teardown of one (no platters), where the voice coil was massive. About the size of a soda can, and cylindrical (the coil alone). The magnet assembly was a cube the size of a stack of CD jewel cases. The die cast frame must've weighed almost a hundred pounds, if memory serves. As a 10yo I was in awe at the sheer amount of engineering that existed in one place.
I remember applying 12v 1A to the voice coil, and it slammed so hard that my mom thought I ran into a wall!
IBM used the term Winchester to describe the 3340, not the 3390 (he is showing the HDA from a 3390). The 3340 had two spindles that each held one 30MB removable storage module. Two 30 MM Drives = 30-30 = Winchester 30-30, the popular lever action rifle that one the west. The 3390, 3380, 3370 and 3350 were not Winchester drives.
Funny fact, in Hungarian language when we speak we call all hard disk drives a "winchester" or "winyó" in general and we even write that in pretty much all non official conversations. Although the more official the situation gets the more likely you will write "merevlemez meghajtó" the literal translation and official word for "hard disk drive"... but even then there are only a few occasions when you would [and then you'd have to] say that (in some highly official situations like interrogation or court i can think of).
The fact that computers were strictly restricted by the CoCom list until 1984 didn't help us to get people into computer science and even after '84 for years a Commodore or a ZX Spectrum was still a big thing for an average mid class family and totally out of reach for the majority of people. The fall of the iron curtain has changed things instantly and a thriving grey market of old replaced rigs mainly from Austrian companies made it somewhat more affordable. In 1990 I've got a non turbo 8088 512K RAM and a [compared to the one in the video) miniature 5.25" size Seagate ST225 20Mb "winyó" ... FFW 30 years and one photo I shoot with a pocket size camera exceed that capacity. :D
The server room I used to work in had a Halon flooding fire system. It was mildly terrifying every Monday morning when they tested the alarms in the building as you wandered if it wasn't by some chance the real thing and you were about to be suffocated.
Except that you wouldn't - It can be disorienting, and that's not a good thing in a fire, but Halon is not and, was never, toxic - that's a myth. It was phased out of use because of it's ozone depletion effect, not because of any safety concerns.
According to wikipedia it is toxic though reasonably mildly, it can 'affect the nervous system and other bodily systems'. Regardless of that it works by displacing oxygen and that means you won't be able to breathe, plus like you say the disorientaion plus the cold, the noise etc. I sure as hell wouldn't want to be in a room when it went off
But I'm no expert, possibly you know more about it than me. I knew it was taken out of use for the reasons you said though.
The standard concentration for fire-fighting applications is in the 5-7% range - leaving plenty of oxygen for breathing - there is no danger of suffocation during a Halon discharge. If you read the Wikipedia article you will have noticed that it specifically states: Exposure to Halon 1301 in the 5 to 7 percent range produces little, if any, noticeable effect. The evacuation is standard fire-drill - there is presumably a fire so you need to go - NOW. There are no requirements to remove existing Halon fire-extinguisher system, and they remain in service in many places, but because Halon gas and Halon equipment is no longer commercially produced, once a system has been discharged, it will need to be migrated to alternative systems.
Sounds reasonable. I don't have the technical knowledge to argue otherwise and the ratio of oxygen only has to be disrupted to a limited amount to achieve a suppression of flames. My comment was only a throwaway one but I maintain I wouldn't want to be in the room when it went off, I'm certain it wouldn't be a pleasant experience. You're probably right that there is no requirement to replace unless it was discharged as thats consistent with many other decomissioning guidelines.
I had a friend in that used to work on these for jc-pennys. They kept all their prices on one of these... They called them Layer-Cakes. The would tear them apart and replace the platters... You would never replace a platter in a mode3rn drive...
Jeremy Catches Yeah, old school. You can buy pallet loads of modern used/faulty drives on ebay.
EEVblog If you want to see some more sweet old school technology, look up "Inside Adam Savage's Cave: Geeking Out about Bits and Bytes" here on TH-cam. He has some interesting items in his collection.
My grandpa used to work on something very similar to these. If I recall correctly, he said they had 12 in. platters and stored 12 MB. I don't know if that was per platter or total. He said that they were designed to have the platters easily removed so that one drive could potentially use more than one set of platters. He also said the magnets on the linear actuators were so powerful that he ruined a very nice mechanical watch buy just putting his wrist near one.
Jeremy Catches You might if it breaks :) But thats quite unlikely and in this day and age someone would just buy a new one so you aren't completely wrong.
+seth094978 Holy shit. Now i'm wondering if the watch could have been degaussed. You know, with those big rings that they used on TV screens (CRTs) when they got magnetized by mistake. They started putting the rings in the sets to degauss them on the start up.
The 3390 Model 1 provided a storage capacity of 3.78 gigabytes to 11.35 gigabytes. The Model 2 had a capacity of 7.56 gigabytes to 22.7 gigabytes.
The 3390 attached via the IBM 3990 Storage Control Models 2 and 3 to all ES/3090 models and IBM 308x models through an additional attachment feature.
First customer deliveries of the 3390 Models 1 and 2 were scheduled for December 1989 at prices ranging from $90,000 to $275,000. A typical configuration consisting of three 3390 Model units was priced at $759,000.
I worked for IBM in the mid-90s in one of their outlets in Germany. The server room was the size of a tennis court, had a 3090 with IIRC 90 gig of HD space. The room was so noisy it was impossible to have a conversation, it was like stading next to a jet at takeoff. Unfortunately I don't remember the details of the mainframe -- I'm a UNIX guy responsible for the RS/6000 machines back then -- but I do remember that the whole floor was shaking from this beast.
LoL, that hard drive is so big that you can see the 1's and 0's on the platters! ;)
Yes, you actually could with fine grain magnetic disclosing film. (I worked on these in the 80's).
by my math it's 8100b/cm², so about 90x90 dots in one square cm, it could be seen with an iPhone camera I bet (with the right lighting), and DEFINITELY a low-powered microscope
Hi
Brings back memories from 1972 I used to work with large water cooled IBM main frames for a bank , 400hz PSU
With loads of removable drives to enable data storage to bunker storage , we move 1tb of spinning storage up from London on weekend 8 Bullins air sprung truck to move it and 10 sparks to couple up the 60amp supplies to each bank we could have used less truck but the insurance value exceeded the value each one could carry . These had large three phase motors belt drive to vertical discs..
Total power for the floor was 1.5 megawatt
I remember those units very well
Thats what we called a Winchesta, mate. in Goleta California they built them, employed alot of folks, Goleta had alot of hard drive industry making r/w heads too. I worked on the big Control Data removable pack drives from the 1970s, some 5 platter, some 10, big and the lid opens like a washing machine, put the plastic round thing with a handle down over the disk assembly, twist and lift it out and add the bottom to it so its sealed. had all the training, we would replace the heads and align them with an alignment pack, have to reach in there with a tool while spinning at 3000 rpm. I worked with 3.5 inch disk certifiers and would work on those in manufacturing and in clean rooms; also had a laser based disk profiler machine I helped develop, it could spin at 10,000 rpm on a glass air bearing spindle holding the disc with vacuum and have to reach in with alignment tools. Some customers would try and mess up and hit the disk? it would destroy the air bearing spindle that cost about $12,000. replacement was listed at 7-10 hours and took special training? I got where I could do the whole thing in under 20 minutes, a robotic assembly aligned to millionths of an inch that scanned top/bot of the disk, air/vacuum and all the electronics, I could replace the spindle and have it running in less than 20 minutes. But the Flash memory? pushed this technology out, company shut down
I really want to go back in time and show them some 128GB SD Cards.
512gb mate
iCloud Henshaw Sorry I mean micro SD's
Deer Viehch what are you guys talking about i thought sd cards only went up to 32 gb because everything says up to 32 gb sd card in the sd card slot
justin rivera no thats just another thing from device to device. I think 128 GB is the biggest you can get, and even that is fuckin perverted.
justin rivera I just researched. The biggest one is 2TB * shoots himself *
EDIT: Ok they are working on these but currently 256 GB is the biggest
I look over at my PC with one of the fastest SSD's in the world inside (Samsung 840 Pro 256GB), but no matter how crazy fast it is, it will still never be as impressive as this hard drive.
One of these days a person will think about the 50,000 terabyte protein memory in their cerebral implant and say that it is not near as impressive as that old SSD from 2014.
Re That's true, but by just opening the two, the 840 Pro will never look as bad-ass as this.
***** It depends on usage. And by then, I won't be using a 256 GB SSD. Hell, they'll probably be ancient hardware in 20 years.
There just is one little detail many forget:
Soon, we won't need more storage for normal gamers/PC-users:
People forget that the data has to be generated somehow and has to have a use. With 40GB/h of film, you have all the data you need - ever. as it would be enough to be at higher resolution than we can see, 3D, 100Hz and multi-language. Bigger films won't be any better - it would be far beyond human limits.
Same for games:
Once you are down to photo-realism, you won't need more storage for textures. Somebody gotta create the game-worlds too, and if we are at 200 different racetracks/Battlefields/Cities, you pretty much have a long time of playing ahead of you. And somebody gotta create those too.
Maybe games will reach 200GB for a 40h game.
But what after that? The sound will be better, than even the most trained person can distinguish, the films better than the best eye can notice, the games long enough to play a year.
At that point, at that won't be that far away as it seems with all the tech-demos of photo-realistic engines, the vast majority won't need any bigger storage units.
***** As Re Said, we are already scratching at human limits.
Take a 75" Screen. to NOTICE any difference, if you have perfect eyesight, you'd have to be at max 3m away.
Likely, you won't sit that close to such a big screen.
If you are any further away, it is virtually impossible to see any difference.
Take 100" and you could go 4m away. Now your whole wall is one gigantic screen.
And for a full movie in that resolution, you need about 80GB.
Take it even further, and lets quadruple the resolution yet again, and we end up with film sizes of 320GB.
I think, with a 200" monitor 3m in front of you, you got everything you ever need.
And that is final.
You now got full 3D, 100Hz, and perfect resolution, together with full surround, multilingual etc.
After that, you only would notice if you got an even bigger screen and moved even closer. And that is something that 99.99% of all people will never do.
14:57 Dude, spin it up! You know you want to. I know you can.
I died a little when you broke the seal and opened it. Then I died more when you started touching the platters.
Did you survive to see how he broke off amagnetic head?
This video was painful to watch for me as well.
How is this video "painful" to watch? It is not a usable(or at least it does not make financial sense to use it) device today and it has little value. Instead of being tossed into a landfill, it served a final purpose by educating the viewer.
"I died a little when you broke the seal and opened it."
You must be new to the EEVblog.
Spocker spock True, but I'm not going to break the wheels off a Rolls Royce Silver Ghost either, even though it is clearly obsolete and has little (functional) value as a car today...
I had to know these (logically) backwards, especially 3380's. Even though today's hardware is far superior, there are layers of abstraction that emulate these virtually still. So it would be a shame if I was asked what one was and could not find an example.
i wish he would have hacked it and seen what was on it
Jenna Love That would have been cool.
Previous to the voice-coil types, they used hydraulics to operate the head actuators. There were valves for distances of 1, 2, 4, 8, etc. Oil leaked over the drive deck and there was a little plastic bottle to catch the runoff.
I took apart some of these old IBM voice-coil types, but with one big coil. The nut holding the platters together was TIGHT! The best part in it was the magnet. The magnet was the size of a building brick and was crazy strong. If you put two of them together, you needed a hammer and wooden blocks to drive them apart. If you put one behind the wall behind someone's CRT monitor, it would warp and distort the pix and drive them nuts.
On H-P drives, they had a similar port, but it wasn't for Halon, it was for air. It would suck some air in through an "absolute" (meaning VERY fine) filter the size of a loaf of bread and blow a bit out the port. You had to measure the air pressure out every week, and when it started downhill, you knew to replace the filter, else you were cruising for a head crash. Halon just doesn't sound right for the IBM drive. There would be no ignition source in there.
The platters for the bigger drives, say 14", made great backplates for nerdy wall clocks.
IBM 2311 hard drive. We used to write our names in the oil on the sysres top platter when we powered the drive off!
Pure IBM quality
Imagine the capacity on that size now
petabytes?
+B0NEZ IBM makes hard drives 120 petabytes. That's 120,000,000 gigabytes. Roughly
+Boyborg690_ oh sorry that's a data repository...
Absolutely not petabytes. I imagine Liam worked it out by multiplying the number of platters, and their size, and comparing that to modern hard drives. Since hard disk density is measured in bits / square centimetre. That and maybe he had a bit of a guess. What are they, 11 inch? Guess how many inches the spindle in the middle takes, and the unused area in the centre. Then do a bit of pi-r-squared, and compare that to modern hard drives, and multiply that by the highest modern drive's capacity.
I dunno if it actually would be slow. Since, the further out in diameter you go, the faster the disk's linear speed is. Seek time might be a bit of a bugger, but that's what buffers are for. You could use an SSD for the buffer, or a few GB of RAM. Price wouldn't be a problem, if we're talking about a hypothetical 11 inch giganto-disk.
As it is, usually HD storage on the petabyte scale uses arrays of 3.5" hard drives. In 1996 , Quantum released their "Bigfoot" series. They were 5 1/4" drives, years after everyone had switched to 3.5" drives. So they had much bigger platters, much higher capacity and a bit more speed. Since PCs had, and still have, 5 1/4" bays (for DVD drives these days, CD-ROM back in 1996), it wasn't a bad idea. In the end though the format disappeared after a while. I wonder why? Bigger diameters have several advantages, and every PC can fit them.
Maybe for archive storage?
I would pay big money for a modern hard drive in this form factor that stored a few dozen terabytes
they already have a 10 TB hard drive..... its the size of a normal 1TB hard drive
so if anything you mean a few HUNDRED TB
***** With technology you always have to decide when go "in". My first Intel-based PC in 1993 cost $3150 and had 33MHz 486, with 4MB of RAM and maybe a 80MB hard drive. But that was a white-box (not brand name) that suited my need to get into Windows and Linux software development. If I had waited say 2 years for something faster or cheaper, I would have lost all that time in developing skills that I used today. And as you rightly say, the gamble that manufacturers make to develop new tech and sell it at the leading edge to early adopters helps develop future products. And the customer is hopefully more competitive, for instance banks might have been able to offer loan approvals in hours instead of days which will make all the difference to their revenue.
Jenna Love
try 1000.
***** either way it would be at least a petabyte
Hi Dave,
Some extra info.
What you call the actuator is in fact the lock to lock the head carriage in the landing zone.
The spring you mentioned is to bring the head carriage back in parking zone in case of a dramatic power failure
The VCM is the thing that controls the location of the heads.
One of the heads (the middle one of each device) is a so called servo-head.
This head is the only head that knows where it is .;-)
The surface of the servo plate is written in the factory and cannot be altered.(Servo head = Read only head)
Very basically it work like this.
Assume heads are on track 0 (Cylinder 0)
Computer want to read data on track 110 of head 7.
The disk will set a difference counter to 110.
Then the VCM is told to move the heads in high speed towards track 110
Every time the servo head passes a servo track the difference counter is subtracted with 1.
When the difference counter is reduced to a certain low number (say 20) the VCM wil start give several full reverse pulses to reduce the speed of the carriage.
When difference counter is (e.g.) 10 the VCM will go in low speed for the last few tracks
The moment the difference counter jumps to 0 the VCM will enter "Track following mode"
That means it will lock itself (electronically) to this track. (We used to push the carriage manually to test the VCM poweramp and see if it could maintain track following)
Head 7 will read the HA (Home Address) of the track and that should match the HA of Track 110
When HA is correct the data will be read. (there are some more records that will be read but that makes it to complicated for this post... and depends if you have a Count Key Data of Fixed block device)
But basically it works like this.
-
B.t.w. Each carriage (VCM) is a standalone device.
So this HDA has 2 devices in them.(in mainframes this is quite normal)
In a mainframe of that time the device addresses works like this.
The data path in mainframes goes like: CPU Channel Control unit Controller Device
1 number for Channel number
1 number for Control unit
1 number for device
So a typical disk address would be 1AE.
That would be channel 1 control unit A and device E
Your HDA would then be addresses 1AE and 1AF
-
Mind you, this is how it worked in the 70's and 80's (time of this HDA)... now it is different.(Among things, larger (= more) addresses)
-
You are correct to assume that the datecode 1991 is because it was an replacement unit.
Might even be that you still can order this device...
Phone IBM and ask if they have a HDA with partnumber 26F7630 in stock. (Or ask a befriended IBM CE to check it... he has the correct phone number for this info)
See FRU sticker at 3:54 (FRU stands for Field Replaceable Unit.)
FRU numbers are the IBM technician part numbers to order a new part.
Well done mate. I wish there was some kind of filter I could apply to the comments on Dave's videos that only shows stuff like this.
I'd love to see the controller for this, someone elsewhere in the comments suggested it was washing machine sized.
The first HD I had was a 40MB Seagate 5.25" drive for which the controller was a full-length ISA card. I think the standard was called ST506. Also around 1991.
+Gavin Ward
3340 and 50 were indeed about washing machine size.
3330, 3380 and 3390 were much larger. (just look on internet for IBM 3330 DASD)
3330 wat the last DASD where the actual disk assembly (not heads!) could be removed by operator.
-
The controller for it just looks like the other disk but has controller logic in it beside the 2 disk addresses.
Keep in mind that there are "Controllers" and "Control Units"
Controllers serve actual disk clusters in the (Hex) range 0-F (It would serve for example 8 of these HDA's Dave showed us.
Control Units serve Controllers and have address ranges like 00-FF (so controllers A0-AF, B0-BF, etc)
Control units only talk to Controllers
-
And then there are Channels.
Channels only talk to Control Units
So the whole data path = CPU - Channel - Control Unit - Controller - Device (one of the 2 moving head carriages Dave showed. Each HDA has actually 2 disks in them!))
Thanks for this, Spike!
Hi Spikejwh1 great 👍
Could well be to order those drives.
I remember that we even had to rebuild scrapped production equipment to produce some more IBM62PC piccolo drives for special customers.
Obviously it didn't matter how much it costs.
does it have a 2 stroke engine on it?
it has a cap powered drill machine ;)
I think it _is_ a two-stroke engine. There's your spark-plug right there: 5:53.
You can see the valve springs at 18:10
It's an 27 speed transmission with racing gaskets
Dem Australian safety boots XD
When I wore those working as a mechanic in a shop, we just called they safety flops
I remember those. I used to work on IBM Midrange systems (the System/36 range of computers). the hard drive platters were 3 feet across and weighed 40 kilos. we usually had one die every 2 years or so. 350MB of storage on the model B and 768MB on the Model D. the actual machine was bigger than a washing machine and made a horrible din as fans and motors ran in it 24 hrs a day. My office had 2 of them and every summer, the air chiller unit would fail and tempretures would shoot up to over 100 degrees. those were the days when computer techs were working at the coal face!
Jeff Barber Time is cruel to old tech. As a 70's Kid. I am still in Awe of this.
Worked in an IBM shop during this period. He's handling a 3390 drive unit. They almost had the same capacity of the 3380s that they replaced, 630Meg, 1260M, or 1890M per actuator/address, for the standard, "D", or "E" models. The 3390s saved 'footprint' space in the data center for the same storage, along with power and noise. A 3380 drive had a sticker on the side, > 70 kgs, recommending 2 CEs to swap them out. An actuator on each side gave the device 2 addresses. Two drives per box, total of 8 boxes per string. The 3390 has two actuators / drive also, as seen in the video. Top & Bottom. If it failed, a single CE could unplug it, slide it out, and replace with a new one. "CE" - IBM Customer Engineer.
That was awesome! Thank you. I design military avionics (for 30+ years), and it was fun to see the engineering behind this beast. Thanks again!
Also, when the drive is spent up, it makes an incredibly good gyro. It was practically impossible to move it when it was at speed.
... is that a fking car engine?
Thats what i thought! XD
Frostbyte It looks like an engine block, in fact, i bet there is enough metal in there to make an engine block. Maybe a V6.
I'd say, more like a 1- or 2-cylinder actually. Source: Renovating several motorcycle engines.
It looks exactly like a lawn mower engine.
+MiGujack3 It scans the grass into digital data.
Pshaw. That's nothing. An ex-boss of mine used to work at Control Data when they were developing a no-holds-barred disk drive. One platter, six feet across, with 8,192 fixed heads. Cost of around $18 million 1978 dollars. Not a problem for their Rodeo-drive customers like the NSA, Westinghouse and Livermore. They were clamoring for disks with low latency, so no head-stepping.
One day he came in to work about midnight, typical hardware hacker time, and the building was almost impossible to enter due to a loud screeching wail sound. The big disk had a head crash, which cascaded from head to head, trashing all the heads and the disk surface. They never had the nerve to try building a second one. It had lasted all of three days and nights.
That strip looks like an aquarium thermometer.
It's an LCD (liquid crystal display) thermometer. Yes, exactly what can be, and is commonly, applied to the glass of an aquarium.
Yup, I had one on my snakearium.
Was rather surprised that EEV had never encountered one prior.
1/4 mil doesn't even come close. These units were used in RAID arrays that had many of these units. What you have was THE field replaceable unit (FRU). Repairs would have been made in a clean lab, probably in Endicott. The heads actually flew by bernoulli effect over the platters and head to disk interference (HDI) was often a terminal event, nevertheless, stuff did happen and you could swap a unit in the raid and the raid would rebuild the data from the other drives. Then send the offending unit back for rebuild. There were also some aeronautical/defense applications in which units similar to these were used. I could tell you more, but I can't.
Is that a motor mount for a Nissan under that thing?
😂 I laughed so hard at this. It literally looks identical to my BDE Nissan 300ZX motor mounts.
Damn, that fly is very well preserved, almost like a time capsule!
That beast would make a great custom case for a gaming rig.
yes
+vector6977 Lol, exactly what I was thinking about!
+vector6977 HehehheEHEHhe
gO tO tRuThCOnteStCOm< ReaD The PREsent
+vector6977 And just a little of engineering for cooling all that
+yomauser Maybe you can use the in/out-Gas Channels for cooling with oil or something like that :D
So many errors and misunderstandings in this. For starters:-
That's just the HDA (Head Disk Assembly) from an IBM 3390, a box the size of a couple of fridges. There were 2 to 6 HDAs (depending on model) in the box, which also contained all the other electronics required to provide the equivalent function to the familiar hard disk. These were introduced in 1989, and the HDAs have absolutely no resemblance whatsoever to the earlier models that they replaced. (I've torn down a number of them myself) This is definitely NOT 1970's technology..
And the HDA cost nothing like $250,000. The whole box, along with the other HDA(s) and electronics, started at $90,000.
that was one awesome video dave. you're the man. hooking it up to a signal generator was hilarious.
I heard the words "bobby dazzler" and knew I had truly come home
I was irritated at the '4Tb in 1.8" = the smallest form factor' statement at first. I was like "Don't we have M.2-SSD's with much more capacity for a bunch of years now?".
Then I looked at the upload date... 😅
This thing in today's market would cost $848,398 AUD, or around $636,298.50 USD.
So basically something JayzTwoCents would put in Little Jay's computer because she wants to get into MSPaint.
Gold!
+lilturk726 $642,858.18 USD *
gabe gabe
Yeah I had it backwards. I put in 848,398 USD instead of AUD.
+lilturk726 made my day.
Or around $40 as a non functioning decoration since that is all it's good for.
At 5MB I don't think that it would've been able to hold MS Paint from Windows 95 but is easily large enough to store all of Jay Z's good songs.
Physically, the case is almost large enough to store all his celebrity endorsements
For me its kind of funny seeing that "old iorn" again. My first day on the job in field service was in 1976 . We were installing "field change orders" on 10 megabyte ,full size disk drives (about the size of a washing machine) The disk drives had a linear bobbin type motor. The field change was adding an RC circuit to the linear motor circuit so that the disk drive would not pickup and play music from the local am radio stations. ...yep ,in some installations it was an issue lol.
Steven Hardy Even in 2015 RF interference is still a nightmare. Open even an Iphone6. Any Metal plates are there only to shield RF Interference.
I worked on these things and the filters actually filtered recirculated air within the unit, not outside air. Additionally, these were were used with mainframe computers like the IBM 360/370 series. Servers as we have today, didn’t exist yet. The platters were double sided except for the top and bottom platter.
Would love to see it connected to a modern pc and running some benches...
Dedicated Player the data rate on that thing is probably way too slow for that
You need to remember that before IDE harddrives, drives were dumb and had their circuitary as part of the controller.The I in IDE stood for Integrated.
I thought only the vacuum tube era was that massive. But this is the transistor era already. Next time I play Quake 1 or watch Alien I will not complain about the computers everywhere wondering what the heck they are for and why they sounds like a factory...
I love this channel. So much enthusiasm.
+brokenSCART JAJAJA
+SmiluHD
Its not a minecraft channel.
+SmiluHD Now that's what i call edgy!
More like a drama queen!
OK 10MB is really nothing nowadays.
But if you make one of these today (same shape and size) with the capacity of 10PB (Petabytes) with a sticker price of $50,000 USD, people would start drooling all over it like they did in the 70s.
Actually they would laugh and stay far far away from it. Single point of failure.
that sensor in the actuator is the "Home" sensor, lets the drive know when the heads are over the "Zero" track.. so the drive knows where to start counting from ..excellent video, haven't seen that type of drive in over 20 years.. please keep these types of video's coming.. btw those heads have more than 1 gap, there is an erase head there is a r/w head, and there is also a dual head that is called a guard head which makes a tunnel on either side of the data channel so adjacent channels do not interfere with each other..
I find it a little sad that now we're all moving to SSD, I mean, possibly the finest piece of mechanical-electric technology you'll ever own, compared to a few chips on a PCB.
6:53 "Otherwise, the halon gas would kill you." I thought this too, then I searched a bit and found this isn't exactly the case. Yes, gases can be dangerous, but not all of them kill datacenter workers.
linagee I don’t actually think he was saying the gas by it’s self would kill you, just that the lack of oxygen once the room is flushed out would.
@@pedroalexander8921 Yes It displaces the air, which is how the fire suppression work. also because of the energy lost to convert the cold liquid gas, it would chill a whole room with the phase change from liquid to gas. just like a CO2 fire extinguisher, the temp drop forms snow ice and dry ice, it's just under enough pressure to keep it liquid, and it occupies a small space so it transports easier.
IBM generally had a controller with the smarts connected to a bunch of not so intelligent drives so I would not expect this item to have much smarts. My first hard drive was from IBM and had 4 Megabytes on a stack of platters wider than this one. Amazing stuff. THESE are the good old days for computers and other hardware.
Also, the top head was frequently a sync track. Lots of little markers for timing, track start, sector start, etc. Not a read/write surface.
"...Genuine hardware bug..." at 13:00 ! Amazing!
That's actually where the term came from.
This Is The Mother Of All Hard Drives :D
Thats nuts thats only 10MB and 4 decades later we can now store 64,000 MB or more on a micro SD card
Samsung and SanDisk have been selling 1TB Micro SD cards for over a year now, so that’s 1,000,000 MB, that’s 100,000 times more memory in something that weighs roughly 82,000 times less. Yes, that’s right, this 10MB hard drive weighs roughly the same weight of 82,000 Micro SD cards, or equivalent to a measly 82 Petabytes!
ahaha, that's not a harddrive...
/THAT'S/ a harddrive.
Not a harddroive 🔪
crocodile dundee of electronics
"Hold the D I'm crocodile done with you..."
You call that a hard drive!??!!
35:40 Goodbye headphone users
The earliest IBM commercial hard drive had something like 50 surfaces (about 26 platters, further separated than the ones in this unit) and ONE head assembly which moved in and out and, while completely out of the platter area, up and down. Next they came up with the multiple head access “comb,” and made the storage media removable, allowing separate data storage for multiple systems (or clients). The most popular ones had 10 or 20 surfaces, and were stored offline in two-part carriers with a handle on top.
But smaller and smaller head gaps made the removable (by the operator between jobs) system too unreliable, especially when people smoked in the computer rooms (an ash particle hit by a flying head was like a rock hit by a lawnmower), so these “head drive assemblies,” or HDAs, were the solution. Each HDA was a separate volume, and by that time, high speed tapes were fast and cheap enough to “swap drives” by copying the current content to a backup tape, then loading another backup tape over it. It took more time than manually swapping removable packs, but not much more, since there was no need to spin the drive down and back up. These modules were replaceable only by the CE (IBM speak for repair tech) and required the entire controller and all volumes to be powered down and the cover removed.
And notice how the access arms move in and out along a radius of the spindle? Not so good for PC sized drivers, because of the noise and vibration (sometimes the big old drives shook like unbalanced top loading washers and driers), and the extra space needed for the fully withdrawn access arms. Today’s access arms swing around a shaft in the corner of the rectangular housing like a phonograph tone arm; the deviation from a straight radial line isn’t large enough to make a difference.
This is such a great piece of history!
Now to think, in 20 years time, we'll have 200 TB harddrives capable of storing 1 government official ego.
The 64th Shadow Nah, you will need at least 1 Terra Quad or more.
The 64th Shadow
You should do some maths the smallest structure we are capable of at the moment are 14 nm , 0.1 nm is roughly a atom.
Fiercesoulking "sructe we a cabale"
What in the fuck are you saying?
The 64th Shadow
Sry I was in a hurry in real we won't have a smaller structure than 1- 0.5 nm because for transistores they make no sense.. The biggest harddrive at the moment is 6TB, so maths say this
14 nm*(1/0.5 nm) * 6TB = roughly 168 TB is the biggest possible HD with in the normal specification of a normal HD in these days
Fiercesoulking you forgot to calculate in 2 dimensions. Shrinking from 14nm to 7nm brings you 4 times as many bits. Shrinking from 14 to 1 brings you 288 times as many bits. And there is still the possibility to create thinner discs and heads to fit more discs into one harddrive case. Even without smaller structures on the disc itself this may also increase capacity. One could also try to store more bits at a single location so that a location is not 0 or 1 but rather 0, 1, 2 or 3 (0, 33, 66 or 100% magnetic force; hdds already use this technique but I'm sure there is still some potential).
So we do a simple calculation: Half the structure size (7 instead of 14nm), twice as many discs and twice as many bits per location would raise the capacity from 6 TB to 96 TB (6TB x 2 x 2 x 2 x 2)
That motor you removed from the bottom, before you could pull out the ACTUATOR assembly is the head LOAD/UNLOAD motor. It was used to remove the heads from the disk before the disks stopped at power down.
If the motor was allowed to come to a stop with the heads in contact with the disk they would STICK.
The LOAD/UNLOAD motor also LOCKS the actuator in the PARK position when the motor is stopped.
does it have a pull start?
bast kahmint
It looks brand new. How did they manage to keep it so clean?
majikmilc
They keep it in a server room, i think.
Zé Bianchi it still works too.
kaptainbastard Did.
mike sixx you would have to ask the new owner that, cause I'd have sold it right away.
I’ve taken apart and repaired one of these. They’re sooooo heavy. We still have the drive platters on our office entrance. I got mine as part of an old system 360. Your drive is a bit newer than the one I had, but it was... 30 years ago. So 🤷♂️.
When that thing spun up, the lights would dim...
Most people do not realise how close the heads floated above the disk surface. If the head was scaled up to be the size of a 747 and the air gap was scaled up accordingly, the 747 would fly into strawberry.
First question is why so many downvotes second question I'd love to see this thing and action it's a beautiful piece of equipment I hope that the disassembly did not destroy it I know that you have to be extra careful with hard drives although I'm no expert
With today's technology, I wonder how much memory you can fit in there.
You mean if you built a RAM-stack and tried fitting it inside? Well - a whole lot!=P
ojkolsrud1 I wonder if it is safe to say a terabyte of RAM can fit in there? I also wonder how much storage you can fit, along with the amount of platters you can fit. It's amazing how technology has changed!
KrazyLemming and still has to. If you look at biology 1TB isn't that much. The human genome, as far as I know, consists of about 3GB of data. Now that is even much when comparing it to flash memory! And when thinking about what one ejaculation consists of: over 100TB of data! WOW! Incredible.
I don't know whether we will ever reach that high data density, but well, if we do, then we appear to have something worth waiting for :)
Maximilian Mustermann 100TB in an ejaculation is a number that often circulates on the internet, but it seems to me that it's only half a human genome (1,5-2GB) of data, just with crazy redundancy..
Bart Guliker Well, if you check the number s: 3.2 billion bases (I don't know the exact english term). The 100-200TB Idea isn't all off. I was far from exact! :)
You really should figure out a way to make this bugger run!!
I love how just the connection port is so big it looks like a slot for an entire modern hard drive
Why didn't you run it and see what was on it.... Why didn't you run benchmarks to see how fast it was?
It's not certain that it's connection port is even IDE, you know. In that case, what you propose would be difficult! But, he should at least tried hooking it up to some power so we could watch it spin up!
ojkolsrud1 it probably would be dangerous because it doesn't have halon gas in it anymore.
hmm, even so, I don't think it would overheat during the period of time needed for it to reach maximum RPM. Still, good point!
Indeed. Probably true.
Its not possible to do so. He would need a System/390 mainframe as this is an IBM drive and would only work with IBM mainframes, the drive controller (this is not an IDE drive so has no onboad controller), a S/390 OS and software capable of running on the mainframe to run the controller and understand the filesystem and file formats.
Basically if he had an S/390 mainframe to put this in it would take up his entire workshop and powering it up would require work too.
I remember when 4MB of RAM cost $10,000
You sound like the guy who used to voice infomercials back in the 90's and 00's. That bit of nostalgia earned my subscription
The original IBM death star drive?
This thing is thicker than your accent
this was maintained pretty nicely, I was amazed to see that all the rubbers were still soft and didn't have any cracks in it
35:40 rip headphone users.
he forgot about micro drives... smallest hdd
Micro drives contain flash memory chips and not magnetic coatings on top of platters... so yeah, entirely different technology.
XmarkedSpot No, microdrives were actual hard disk drives with a 1 inch platter. Same case as a CF card (thick version), but back then when they were a thing they had a much larger capacity than one could have at a reasonable price on Flash-based CFs.
kilrahvp
You're actually right!
I even forgot that i own one of those, sheesh i mixed it up, my bad.
XmarkedSpot I bought one a good 10 years ago. Was fine even if rather slow, and it did what hard drives like to do best and failed on me after only a few months. Can't remember what I was using it in though.
At least I got to take it apart and marvel for a while at the impressively tiny machinery... quite the opposite of the one in the video, lol.
kilrahvp
I got mine with yet unseen 2GB for my first-gen digital camera (~2003 Canon, capturing a whopping 3.1 megapixels).
The drive actually outlived the camera (8yrs of heavy use) and who knows, probably still might work. I guess it would have failed me if i wouldn't have a backup routine ;)
Having a 64GB sdxc by now doesn't really make me miss it... but i agree, the technology still remains very impressive.
For the record 250,000 USD at the time of the drive's production is around 800,000 current USD. If inflation wasn't already calculated, then this is a better number for scale perception. In other words, that's really expensive!!!
At the cost rate of that unit, a 1 gigabyte hard drive array in that day, would have cost: 25 Billion Dollars!
At the cost rate of that unit, a 1 terabyte hard drive array in that day, would have cost : 2.500000000×10¹³ Dollars!
W O W !
holy fucking shit lol
5argeTech they way I calculated it, a modern 3TB HDD would cost 75 Billion Dollars based on 10mb=250,000.00
swifty1969 That's is what my answer is on a calculator of its day! :) look at the answer it is 75 billion!
This hard drive was $250,000 and could store 1.89/3.78 GB, so a 1GB drive would have costed anything between $66,000 and $132,000.
So a 1TB drive would have costed anything between $66,137,500 and $132,275,000.
cool! but poor hard drive
Wow, takes me back. Used to work on GE Medical CAT scanners (circa 1981) that stored scan data on CDC 96MB removable platters, the entire drive enclosure was about the size of a small washing machine. Replacing a read/write head in the field was a treat; no clean room, basic hand tools and an old school Tektronix storage oscilloscope. Was only required to do it twice but I can say the installation and calibration instructions provided by CDC were outstanding (provided on Microfiche slides)!
He's the Steve Irwin of computers.
Crikey, look at the discs on that one.
If thats the HDD then how big is the PC ?
Are we in the PC right now? :D
TheDiamondHack Yeah I would assume as much. Just being funny :3
AlchemicTempest It was as big as a large room.
I have an unexplainable but very real soft spot for old hard-drives the way some people do for puppies and things. I just love them!
Plus those mid 1990s hdds have the best magnets to re-purpose to keep even he heaviest grocery list stuck to the fridge. Hehe.
"pornographically smooth"
Four years later, we have 8TB Hard Drives and 3.8TB SSDs
And 2TB usb flash drives
Flash Drives and SSDs are pretty much the same thing...
Problem with SSDs is the slow write speed and that they wear out.
The wear problem is why SSDs are not used in servers.
That is no longer the case.
If there would be willing customers, there'd be much bigger SSDs in 2.5" or 3.5" size.
I worked as a student employee at IBM in Rochester, Minnesota, USA between Oct. 1989 and Aug. 1990. I worked in the platter manufacturing area. The section I was in produced 3.5 inch and 5.25 inch platters.One of the techs on in the coating area told me that one of the 5.25 platters would hold up 125 Meg of data. I was amazed because at that time I wanted to get an IBM desktop which the largest hard drive available was around 60 Meg. Another department produced the 11 inch and 13 inch platters. You estimate how much the larger platters would hold based off of what I have described. I was very cool working there because I got to see how they were made from blanks all the way through to final test. The final test area had about 10 quad platter testers. In that area we would have to change out the heads on the testers. We would inspect them under a microscope and clean them. A friend of mine worked in another area when they made the heads. They would manufacture a long block and laser cut the heads from that. Very cool experience!
Am I the only one who can here the 15KHZ?
Average human hearing goes up to 20kHz, so no, you probably aren't
Karrotz Witte im 47 and i could just hear it.
***** Human hearing tends to decay as one ages.
yes that is true, but some people cant hear over 15kHz at 30 something. I have looked after my hearing, cant say about the eyesight though.
IVoidwarrantys More likely its signal distortion from the cheap headphones/ pc speakers you are using.
why is he talking like that?
He is happy i guess
BTW the thermometer strip is a standard-looking LCD thermometer. The numbers are printed underneath, and each segment of the LCD becomes transparent / opaque at particular temperatures, that's how it works.
For the heads, I think the reason there's 4 contacts, with only 2 being used, is so they can use the same heads upside-down, which would reverse the L / R direction. Saves them having to make and use 2 types otherwise.
If that's just 10MB, then the data density on the discs is going to be pretty close to that of early '80s floppy drives. Which wasn't much!
I think pc hardware barely evolved in the last 7 years...cpu are only 20-25% more powerfull,4-8gb of ram is still standard,ddr3 is still here,hard drive capacity went from 320-500gb to 500gb-1tb,we still use Sata and besides graphic cards getting a huge boost,nothing else dramatically changed...
That's exactly what I was thinking. Even two years after your comment, almost nothing changed. Sure, they got a bit faster and stuff, but nothing exciting, just CPU bugs :D
Change has been slowing down for sure, but 8GB is more common than 4 now and were finally starting to see systems with 32GB of RAM.
How many petabytes could a hard drive that size be today?
Well, if you have a lot of money you can try with msata 1Tb SSDs, they are 51 x 30 x 4mm big so multiply that by 1000, you get a petabyte of crazy fast storage for about 250 000€ in a 32*60*32cm box
Yeah, pretty sick tech that we have today, you could probably fit 1 of these petabyte boxes in that bulky IBM HDD enclosure.
Also, considering that the internet 1.2 Zetabytes big, you would need 1.2 Million of these boxes units. It would be about 0,06144 cubic meters per enclosure, multiply that by 1 200 000 that makes 73 728 cubic meters,
A 747 has a cargo + passenger volume of about 1000 cubic meters so that's 73 Boeing 747 to transport the entire internet
And AirFrance has exactly 73 of these jumbo jets...
It would cost 1 200 000*250 000 = 300 B€ of SSDs and it would weight about 10 200 tons of SSDs, and Shell can have this amount of money...
After all of this math, the conclusion is:
AirFrance and Shell are going to fuse together to have enough money to buy all of the SSD's they need to put them in the AirFrance's jumbo jet fleet, send them to Australia and then they will use EMP bombs to destroy the internet and keep their copy of the internet in hostage to get anything they want in return.
My brain hurts so bad
a crapload
I managed a datacenter for 15 years and the halon was no joke, we had training and were told if the alarms went off to vacate the room asap, injury could occur as it "replaces" the oxygen in the room.
Hey guess what "most expensive hard drive"?
We've got SSD's bitch.
Gzpy butt ssd is not a hdd, ssd = solid state drive. Hdd = hard disk drive
THAT'S A HARD DRIVE JESUS CHRIST THE BEAST LOOKS LIKE A RADIATOR OR A REFRIGERATOR BUT NOT A HARD DRIVE.
Looks like an IBM 3970, and there would be a room full of these for banks, etc., you provided an excellent But they were built for super reliability, and IBM built them like tanks. I used to program mainframes and super-computers, and this was certainly a pleasant walk down memory lane! Thanks again !!
When he said, "Hi - welcome to teardown...Tuesday", it reminded me of PeeWee Herman. And how 'bout a testosterone shot to get the pitch of your voice down into the "man" range...
Actually he is a shoo-in for Dick Simnel if Raising Steam ever gets the Tom Hanks/Spielberg treatment. All that and a top tenor and you want it treated with chemicals?
Silly man.
speak normal you don't have to stop and pause after every word
that's his style, don't like it, don't watch, no one is forcing you to do it
666Tomato666 Shut the fuck up and stop trying to ignore the fact that there are people who don't like what you do
cornstorch riiiight, because I should change everybody around me in my image
great idea! *everybody* should do it! /s
666Tomato666 That's literally the exact opposite of what I sad to do
I told you to change the way you react to it
potch paul He can speak however he like, if this is how he speaks then this is how he speaks. I for one find his speaking incredibly pedagogic.
I expected you to pull off a "That's not a hard drive, this is a hard drive"
Omg!! His voice!!! SOO ANOYING ! :O
I was going to type the same thing but I honestly didn't want to be mean or seem like a troll! Nothing a little testosterone couldn't fix, that or voice training.
Artemus Rodricq I don't want to be a troll, I just said that his voice irrited me...