ฝัง
- เผยแพร่เมื่อ 5 ก.ย. 2024
- Stacking some Intel Express 510T Switches and 220T hubs with special cables to combine multiple physical units into logical ones for easier management. All of this can be done with the Intel Device View software running on Windows XP. We'll also try to get a 1gigabit fiber link going to a Cisco 7200 VXR router.
Interview with Doug Boom: • The Ethernet Age: Inte...
Previous Intel Express Video: • Stackable '90s Intel N...
Intel Device View: archive.org/de...
Check me out on Patreon: / clabretro
Music by Karl Casey @ White Bat Audio
#intel #networking #retrotech
Rack stuff
StarTech 25U Rack: amzn.to/3mEB7hS
Tripp Lite SMART1500LCD UPS: amzn.to/3KZW3Jw
1U 24 Port Patch Panel: amzn.to/3Nm0bFa
1U Brush Panel: amzn.to/3mExAA3
1U Rack Shelf: amzn.to/3oaDclT
Video gear
Camera: amzn.to/4al3xjA
Main mics: amzn.to/4dCUuO2
Desk mic: amzn.to/3ye8BsV
Note: The above are Amazon affiliate links. It doesn't cost you extra, but I'll receive a commission which will help keep the content coming. I only link to things I've personally ordered.
![ตีสิบเดย์ [FULL] | ตัวตึง! ปากกล้า! ท้ารบ! "ทนายไพศาล เรืองฤทธิ์"](http://i.ytimg.com/vi/pU3dXTsaBVc/mqdefault.jpg)
A patron named Luke pointed out I need to switch the media-type of the gigabit interface on the 7200 to gbic (that's why the RJ45 en light was always on). Sure enough -- both link lights on the fiber modules are on now. Still more troubleshooting to do, but fiber videos coming in the future!
Another way to test is to loopback the fiber. tx to rx on the same port with one strand/one half of the pair.
You’re welcome!! I love these adventures!
On those short distances, cables like single-mode (smf) or multi-mode (mmf) don't really matter. You should at least get a link. What does matter is the light color. Usually for mm it is 850nm and for sm 1310nm. Those can't be mixed. Also if You have LX/LR modules (long-range), You should at least use a 3m fiber to connect. Otherwise things can be damaged.
A nibble is four bits... Early Fiber was a nightmare. h\Having to come into an older network and try and work with it was a nightmare job. SFP and QSFP simplified the process and it works just as if it were cat5/6/8.
@InconsistentManner I thought a nibble was two bits and a word was four bits. Where did I go wrong in life?
“When she sees my stacks, I bet you that she calls.” Now I finally understand what these lyrics mean.
Old enterprise software is so fascinating, it feels way more polished and finished than modern software that always feels like a beta at best.
I'd rather use it than Aruba Central 😭
I have some older stuff at home, and it sure feels like someone cared compared to the new "fancy" BS I have to deal with at work.
Because it was better.
The old network equipment was made to last. The new absolutely doesn't
As a meraki user. Yep
The reason for bits is historic: a 'byte' wasn't always define as 8 bits. Depending on the machine and links involved a 'byte' would refer to the number of bits needed to represent a character. 6 and 9 bits for one byte wasn't uncommon in the beginning of computing, and many other numbers of bits were used. As a result, you couldn't define the bandwidth of a communication channel by bytes/s without also throwing in how many bits were in a byte. As a result, using bits/s was the only universally understood figure, so it stuck.
As time progressed 8bits = 1byte became pretty universal, but at that point communication systems were used to be defined in bits/s, so that's why we have it today.
Also, communications theory is all concerned with single bits. Go back and read Shannon 1948 and it's all about symbols, which are bits. Stacking them together into higher layers is a job for a protocol, not a modulation scheme.
The 850nm modules are multi-mode transceivers. The 1310nm modules are single-mode transceivers. Multi-mode is for shorter distances, usually up to 300 meters. They do come in a couple of variants, the OM2+ versions usually have an inner core size of 50um, while the OM1 usually is 62.5um. The way the different multimode patch cables are differentiated is through their colour, gray is legacy OM1, modern OM1 and OM2 are orange, the cyan one is OM3 and so on, there is plenty of reference material to be found if you wanna geek out over the details. Yellow is usually used only for single-mode fiber, which has an inner core of 9um. I have no idea why multi-mode is still thriving even today in enterprise DC networks cause there is no real price difference today, though it used to be huge differences in the past, hence the popularity of multi-mode in the past.
All that said, you are right about the receive and transmit, usually it is indicated by small arrows on the transceiver themselves, and by simply switching them around, you should get link, but as you pointed out you did not. There is two likely reasons, one the interface is administratively shut down on the router or switch side or both, or it is simply that the modules are horribly dirty (clean them out with a lint-free tiny q-tip style thing with isoprop, do several cleans to make sure - usually if they are kept in very dirty environments with their dustcaps off, an oily film can build up on optical surface of the connectors themselves).
Another reason might be that on the router side you need to configure the router to use the GBIC rather than the copper interface. I remember buying the G1 controller card when it was brand new and all the rage for a measly $10k. But I don't remember if I had to set a particular config flag to indicated optical rather than electrical interface. Hope this helps!
I think OM3 is the oldest you can still buy at fiber stores. Wdym modern OM1/2
There were Cisco LH GBICs (1310nm) that could use Multimode with a "launch patch cable", and almost all optics will work to some degree on multimode, I believe, over very short distances.
@@thewhitefalcon8539 Like from the last couple of decades, before that I only saw gray multimode.
The manual for the Intel 1000SX module lists 50 µm or 62.5 µm MMF cables and uses 1261nm < λ < 1360nm wavelength. Cisco has GBIC-LX/LH (WS-G5486) that operates at a compatible wavelength and can use either SM or MM fiber but being a LX optic the laser intensity can be too high for connections less than a few 10's of meters causing errors on a SX type optic in the Intel module.
@@CharlesLaCour That's the first that I've ever heard of an SX style optic using a wavelength other than 850nm, And I've not seen interoperability between 850nm and 1310nm optics. But I guess anything is possible.
You NEED a wall of these. You need it yesterday. Give in to the stack!
Now we just need to combine the Intel Stack with the Linksys Stack. And perhaps we need to make a Cisco Stack as well.
Nortel Stack soon?
@@the-perfidious Only if you get the ring tone too.
I have some Cisco small business switches that were released early into the Linksys acquisition era. Not awful, but leaves some wanting.
Need to get some 3com too
Device view uses SNMP communities to manage the switch, bypassing the username and password.
As long as the SNMP write community is default "public" you can do the reset and whatever you want.
For cleaning fiber ports, I highly recommend the click style (sort of a pen shape, a cap with a lanyard). You will need one for SC connections (the square ones) as well as LC.(the SFP ones). They can clean ports and cables. Two clicks and you're clean. A must have!
Its been years but I still remember the distinct click of these when I did datacenter work...
Yes, and there is a great risk of damaging the ports if you plug in without cleaning first. And then you need a special microscope to verify the cleaning process and specific training to clean and interpret the result in the microscope!
The real fun in all of this back in the day,
was doing it all live on a client's property, with a client who didn't follow all the steps arranged by the designers of your company..
They would not make a scheduled arrangement to be ready and chaos would ensue with people breathing down your neck throughout the entire process..
In other words, they didn't understand why they need to put out word to all their customer base that they would be shut down for an upgrade or maintenance work..
which meant they could be losing a lot of money and customer problems during the entire process,
.. and it was all YOUR FAULT.. because YOU are the guy on site.
So you're basically everyone's punching bag.
And to add to all this fun, dealing with a spider web if patch cords done for a couple years by an IT team who rarely labeled anything very well, and definitely never learned to or really caree about cabling
It's still like that, mostly. you have a bit more leverage regarding being the blame guy. Though they'll still be unprepared for absolutely everything unless their management is competent and would confront them for sloppiness from the start. But usually there's almost no hope and probably still gonna be like it in 20 years. Different tech bit but same attitude choice th
at people make for good or bad.
Pile of networking, exciting Sunday
You can easily test if fiber ports are good by just doing a loopback test, that is connecting tx to rx with a single fiber. If the port is good, link will be up. And splitting the bundled tx+rx is common practice, no need to worry.
I usually just use 10x for bytes to bits rather than 8 to allow for some frame overhead etc. I agree that using bits instead of bytes was a marketing thing, same as hard drives using 1000 vs 1024 for a kilobyte. Back in the day bridges were used to join hubs, which eventually turned into boxes full of bridges which came to be called switches. Thanks for the video!
the port test on the hubs at 24:03 was way too satisfying
It's kinda scary if someone gets a copy of intel device view and gets on the network they could factory reset your hubs and switches without authenticating
yeah... that was wild
This takes me back 20 years. I seem to recall them being popular with auto shops.
i want an autoshop
Maybe because they don't have any fans
All of these peices stacked up together remind me of stereo component stacks in the 80s and 90s. Love it, haha
I love your content. You are the Techmoan of enterprise computing. Keep it coming!
thank you!
I really like your layed back conversation. I'm currently learning Cisco for my CCST exam in the (hopefully) near future. and then go on to CCNA/CCNP Voice. Already have like 40U of Cisco devices collected.
You als mentioned The Serial on one of your other videos which I also watch. As I'm 46 I know the modem era with US Robotics 56k6 modem annoying mom on the land line going on forever on Westwood Chat. These were fun times.
Thank you, and good luck on your exams!
I'll join you on your fibre journey, I've finally bit the bullet and bought stuff to play around with the mysterious SFP ports in my rack.
nice!
I've jumped into SFP+ myself. For short runs, you can use DAC instead of fibre, which is what I've done until I migrate my closet. From what I understand, fibre can picky with the hardware you use.
@@SoberAddiction I'm doing a raspberry pi 2.5g home network router because I got FTTP recently and all I've got to say is "well that escalated quickly" how do you people afford this stuff? Why do I need to put the thing in the thing and then plug it into the thing? Why is that not flashing?
I finally found something more expensive than warhammer.
If you need to know the wire type for an optic, check the nm (nanometer) value, 1310 usually means single mode, 850 means multi mode. Both sides needs to be the same to not throw errors so your cisco and intel optics must have the same number.
Comms refers to bits rather than bytes because, for the most part, data is moved a single bit at a time. Over copper, you need to reach 1000BaseT before you leave the realm of transmitting (or receiving) one-bit-at-a-time. You are right about Marketing though. For decades we in IT used kilo- and mega- prefixes to refer to powers of 2. Then Marketing figures that they could sell a 110GB drive as "120GB". I'm surprised that IBM didn't describe their 5.25" disks as "368 kilobytes".
You just described the difference between KBytes/KiBiBytes and MBytes/MibiBytes, Gbytes/GiBibytes. BiBytes denotes the 10^x notation, Bytes the 2^x notation. But you're right, the manufaturer aren't being fair in their notations, selling a 120 GiBibyte drive as a 120Gbyte drive.
@@paulmichaelfreedman8334 The drives are as many gigabytes as they say they are. Linux and macOS show this correctly. Windows still uses the old JEDEC standard (which is deprecated) and shows it wrong. 'The definitions of kilo, giga, and mega based on powers of two are included only to reflect common usage. IEEE/ASTM SI 10-1997 states "This practice frequently leads to confusion and is deprecated."'
@@bindkeys Windows uses the 2^x notation so 1 GB is actually 1024 MB, and 1 MB is 1024 KB. But is always also shows the free space in single Bytes
Who's more important when it comes to computers, the ignorant morons who won't ever understand it, or the technicians who have taught themselves to work with binary?
Excited for your Fiber journey.
Bits are a measure of speed.
Bytes are a measure of space/stroage.
Nice stacking! I used to manage the early HP hubs. Similar management, but not as colorful!
Great work.
It would be interesting to sniff at what code(s) are sent to do a factory reset on the web interface - also, that's nutty that it can be done "remotely" (through HTTP), but not locally.
agree! one day I'll watch the snmp traffic.
@@clabretro: Or use Wireshark. I'll be switching (no pun intended) to some Cisco Catalyst equipment I've had stored away, and was amazed for how easily they can be reset locally and initially configured through a web interface. But I'm looking for the "Cisco Network Assistant" software.
Random thing i learnt on bits vs bytes
In networking you usually care more about the transfer rate (transfers per second) or bits which is usually how alot of the data is managed/views on a base level within the switching/routing.
On the "client" or "endpoint" side you really only care about the data means together
This was more of a actual reason in the early days and stuck around due to the good ole tradition.
That reference to vxHumvee gives me vibes that the management module might be running vxWorks as it's OS.
I bet Humvee was the internal name of a development board or chipset.
You can have equipment working on byte sizes other than 8 bits. That's why we do it in bits instead of bytes. Nothing to do with marketing.
Nope. Ethernet is a serialize system transmitting one bit at a time. Thus "bits per second". Serial is the same. (not to get into the modem "baud" debate.)
rare though?
@@peppigue Depends on the field. In embedded systems and real-time environments it's quite common.
@@DenDodde cool, anything i can take a look at?
@@peppigue AS15531 aka MIL-STD-1553 for example use a 20bit byte.
Enterprise networking gear (especially old stuff) is notorious to not fully leading on how much of the ecosystem you need to buy into. It looks like you can just dip your toe in, then you find out you need x to do y over and over and over again, until you’ve spent 3x than you were initially prepared to spend :)
I'm not sure if this is the actual reason for the kilobit/megabit vs kilobyte/megabyte for network speed, but networks talk about transmission rates, i.e. data per-unit time, whereas computers deal volumes of data, i.e. words of data.
In terms of the OSI model, the transmission rate comes from layer 1, the physical layer. It has no concept of bytes because as far as it's concerned, the bits come in, it encodes them into an electrical signal, optical signal, or radio signal, and transmits them at a given rate. Layer 2, the data link layer, is the first place we see bytes, and that's relating to how much data we can transmit at a time, or the size of the frame, known as the Maximum Transmission Unit or MTU.
It's no coincidence that it's also the first time we talk about a volume of data, which is always measured in bytes. As we all know, bytes are always counted in powers of two, i.e. 2^10 = 1024 or 1KB (SI can f**k right off, the IEC definition, and the IEC itself, came first!), and volumes are always discussed in bytes, KB, MB, etc.. Whereas speed is always discussed in terms of transmission rates, i.e. bits per-second, which isn't counted in powers of two, but powers of ten, i.e. 1Kbps != 2^10, it's equal to 10^3.
In other words, it's the difference between the transmission rate, and the volume of data to be transmitted.
This is exciting! Your sneak peaks on Instagram are great! 😂
I wonder if there are any Intel modules that have Gigabit SFP or GBIC ports 🤔
Also, have you perhaps cleaned the fiber cables? Unfortunately the optics on the Intel side are baked in... Perhaps is there an option to no shut (enable) it in the Intel software?
@23:50 I used to have a 13 slot Cisco 5500 that had 12 slots filled, we always called it the Christmas tree and it was fun to watch on the very rare occasion it had to power cycle. Best most robust switch I ever managed too, I was sad when it had to go due to voip.
@ 24:20
Your like me, id buy them just to put them on a shelf in my room to look at.
I'd have like a whole little (or big) tower of power going on lol.
It's definitely cool to see this older tech I'd otherwise never get to see.
Subbed. Love your enthusiasm in general and especially your interest in old (but still fascinating!) technology.
thank you!
Glad to hear You see the light! Love your retro videos.
Always interesting to see how the meta of how this networking equipment evolved from his type of stuff to modern SFP based modules
Yeah, I think I see an issue.
The GBIC module you inserted is labeled as being rated for 850nm. If I look on the Intel 1000sx module datasheet, it uses as 1300nm laser. Your cable works for both 850nm and 1300nm though. Do you have a GBIC for 1300nm?
Out of curiosity do you have a link to/lead on the module datasheet? I'm not aware of an 850nm 1000BASE-SX variant; that's 1000BASE-LX. I couldn't find an explicit confirmation of the wavelength, only that the SX module takes 50µ/62.5µ MMF.
@@georgemachappyThe 850nm on GBIC is written on the modules. You can see it on 28:28 or 30:39. I got the 1300nm from the 1000SX module datasheet which I found a scanned copy of, and it would make sense to me given the problem description, and would confirm with what you said, that there is no 850nm 1000base-SX.
That OM3 cable ended up being just fine, oddly enough there was a configuration issue on the Cisco side (I never actually set the interface media-type to gbic) and you actually need to plug in a *third* stack cable into the Intel module and connect it up to the matrix module as well!
Great stuff brother. It's super cool to see you check out these old network components. Also, I've noticed you're accumulating a whole lot "new" rack-mountable network gear. Any plans to add a second rack and REALLY get a stack going? Love your stuff dude. Keep it going!
Thank you! And yup... planning out a taller rack now.
Damn dude you're stacked! Intel stacks, Linksys stacks, Cisco stacks. Need more Sun stacks tho.
Another exciting pile of late 90s hardware :D
Thanks for explaining the difference between hubs and switches.
Bytes are usually used for block storage. Networking has always used bits for data transfer. Great vids. Wish I still had my Intel 550 8-port “layer 2/3” switch to send to you.
This looks like a really great solution...in need of a problem.
Weird little shortcut you may find handy with Cisco IOS. When you're in conf t changing attributes of an interface, you don't actually have to exit to switch to another interface. Example, you can do:
conf t
int gi1/0/8
switchport access vlan 1003
shut
no shut
int gi1/0/10
switchport access vlan 1004
shut
no shut
exit
Doesn't save you a TON of time, unless you're configuring more than a few interfaces at a time.
I didn't know that!
@@clabretro Some of the newer IOSes let you use a range command. Like
int range gi1/0/1 - 48 and that would select all 48 ports on a single switch in a stack. Int range also lets you select ports separated by commas up to 8 at a time.
int range gi1/0/2, gi1/0/3, gi2/0/48, gi4/0/23
If your cisco switch is running NX-OS (used by Nexus series switches) you leave off the word range for some reason.
Everything in NX-OS is "the same, but different" than IOS. Maybe 85% the same, but that extra 15% is what gets you.
It would be like alternating between Linux and Solaris.
@@stonent yessir! Also another one of the things that pisses me off about Cisco IOS. No two pliances seem to ever have an identical command set.
Example, I've got 53 access layer switches throughout the company we've replaced. First step is sh int link, see what hasn't had a connection in months, yeet that trash before even starting. Saul Goodman. Done.
Now, we've also got a bunch of switches in the data center for servers. Which have a ton of patches dating back fifteen years, some of them dark. Can ya do a sh int link on that? Naw. The 6500 chassis doesn't support that.
I used to run large LANs with a few friends, using 6 of the 10/100 switches. Was really nice! Worked really well, but eventually we swtiched up to 1Gbit and HP switches.
Yea the price of old stuff keeps me from messing around with my home lab. In some cases, I had to wait years before something would show up. Then I can't argue on the fiber. I picked up 3 Cisco switches. I then ordered some SFP modules. Used the instructions to override the Cisco only SFP module. That did not work. After snooping around e-bay. I found the correct modules. It also worked in the cisco router.
I managed an Intel factory that built these products back around 2000-2001
amazing!
wish I had something interesting to ask but I'm just in awe. these units seem really well built.
Not saving to flash reminds me of a previous employment where we had repeat calls out to troubleshoot our devices at a particular location under construction. Almost every time, my line of questioning was "Did you lose power or disconnect your switches between our last visit and now? Can you check the port configs for all our devices that go back to X/Y/Z closet? They're back at defaults? I see... I have a suggestion for your network administrator."
@21:16 "oh we're so stacked" phrases you can use in any context for 500 alex
For the Bits vs Bytes discussion: If you go the route of marketing being responsible for that, its probably not to make it sound more, but just to protect themselves from the average user who doesn't understand protocol overhead, so for them its just "some other unit" and not "they are scamming me, because I only get about 110 MB/s through my 125 MB/s (so gigabit) link", since they usually only observe payload data (e.g. during file transfers).
4:50 - it's got an Intel i9! lol
Fibre takes quite a but to get it right even on modern gear most vendors require their branded SFPs and then the Single Mode or Multimode and the Wavelength to contend with, the SX module should be a Multimode and LX would be the Single Mode, the cable you have OM3 should be fine for multimode as it's a higher spec then OM1 and backwards compatible. Should be able to see if you are getting TX with a camera as it's similar to infrared light most multimode modules should be 850nm.
Networks are platform agnostic. Bytes used to not be the defacto standard size. Some computer platforms used to use different word sizes. I remember watching a video of a compute platform of something like 40 bits in a word. Networks are also not computing but sending strings of bits. Being able to send a billion bits per second means you're sending a gigabit per second.
Such an interesting and well made video
Also stacked obviously
I don't know if you figured out the fiber interconnect problem or not, but I saw another comment mentioning the cleaning click. Definitely do that. The other thing that I can think is maybe it is a differential between the connectors and the fiber jacks. I don't have historical stuff about when it came into play, but there are SPC and APC type connectors and jacks. SPC is a direct connect, sort of like putting two pencil eraser heads together. APC is angled. You have to look really close, but you can see that there is a slight angle on the extensions that come out of the connector. The inside of the jack is also angled. They need to fit together so that the light can follow the path with least reflection. APC is specialized for minimizing reflection. I don't have exact numbers on how much better it is, but it is what it is. The point being that the jack and the connectors have to be the same. APC or SPC. Then of course you've already looked at the transmit and receive issue. Somebody else already made a comment about the wavelengths and compatibility between the transceivers. I would think it should be in the documentation what type of cable, single or multimode, and what type of connector should be used. But these are ancient and you already said the documentation is sparse. Can't even imagine the appropriate documentation for the appropriate version of the fiber modules. I'm typing too much, but hopefully somebody else didn't mention SPC and APC. They may have. I just can't read through all the comments as many as there are. Good luck!
it´s not SPC ! it is SC-UPC which means Square Connector - Ultra Physical Contact. ps. always use APC when you are doing shorter runs within a building with LC or SC (Lucent Connector or Square Connector).
The difference for Bits vs Bytes is Throughput vs Storage. Throughput is always represented in Bits per Second.
The reason that networks are measured in bits rather than bytes is that they are serial in nature - they transmit bit by bit only - this was the case when using analogue lines where we measured in baud (which relates to how the signal changes each second and is what determines how many bits can be passed each second). Bytes are a higher level construct and although we think of a bytes as 8 bits, not all communications use 8 bit bytes - e.g. 7 bit bytes were used when transmitting text which is why uuencode/base64 became a thing as if you wanted to send data that used 8 bit bytes you needed to encode the data into 7 bit bytes. In network world you see octets as a substitute for byte - an octet is always 8 bits and so avoids ambiguity.
@clabretro you need to use 62um fiber cable for sx connections and it is multimode and usually orange. Aqua is 50 micron multimode and yellow is 9 micron single mode.
They can be reset from the intel device view because its hard to find lol very secure.
yes there are eight bits in a byte, for filesystems and data processing. A lot of the time network traffic is measured in metric, ten bits, because it cares about the rate of flow rather that storing it (also similar to the way hard drive storage is sold/marketed in raw bits)
i really like the lcd screen for some reason; makes it more retro somehow
That's what these Matrix modules are for. Now I know. Thanks
I'm pretty sure network speeds are measured in bits because data is transmitted serially. I imagine the first network engineers that coined it were thinking of network speeds similar to baud rates. Regardless, cool network stack! It's neat seeing the different ways network brands were doing this kind of thing. Hope to see some large network with both the Intel stack and the Cisco Powerstack.
Also check the cables, with those old SX optics you'll likely need a mode conditioning OM1 patch cord (62.5/125) for a run that short. Check out Cisco's installation note for the WS-G5484 for more info. I'm not sure what Inte's side would require, but I dealt with tons of those 1000Base-SX optics in 3550s back in the day.
From intel: Notes on connecting hubs running at 100 Mbps 1 Use only Intel Cascade Cable (product code EE110CC) to stack hubs.
2 Never connect hubs operating at 100 Mbps with TPE cable unless you use an Ethernet Module.
3 No configuration of the speed or duplex mode is needed.
4 Do not change the duplex mode to full-duplex
Oh we are SO stacked
i have a 510 t with the fibre module and i used to to connect to a netgear swicth with a fibre modlue it worked well at the time and this was very nostalgic for me .
Networking uses bits because, historically, almost everything was serial. So, the simplistic answer is: bits per second for serial buses, bytes per seconds for parallel ones.
Also, to nitpick even more 😂... the "proper" term for 8 bits is octet. Early architectures even used bytes with bit lengths other than 8... but 8 bits to a byte is a more "modern" de facto standard.
Don’t forget the good old ‘nibble’ of 4 bits! 😂
"I don't care what they say, 36-bits is here to stay."
Turning the whole double-stack on at once drew so much onrush current that it made *my* lights dim for a second
When networks first started appearing not every computer system used 8 bit words as a byte. They were expressed as word lengths and could be from 1 bit to more than 18 bit words. A Byte also does not fully express the word length due to error correction, headers, framing etc
That looks like it would be more of an edge switch than a DC switch, messy cables are the default in closets. For the fiber connection double check the auto-negotiation and flow control settings for the interfaces between both devices match as the defaults may not. Generally you want to match the yellow (single mode) fiber with the LX gbics and the aqua (multimode) fiber with SX gbics.
You might want to take a look at Cisco's T-Rex traffic generator project. It's not super user friendly, but it can generate pretty much anything you desire at rates that will have no problem overwhelming this old gear.
I've been eyeing that, might be what I go with
When you talk about networking/transfer speed, you always use bits. When you talk about storage space / file size, you use Bytes.
;)
I've been busy all day, so I'm late, but I bet this is as exciting as I imagine! +1 for Intel using db9 rather than rj45 things. at least it's "common" (unlike before the 90s)
networks can only (in general) push one bit down a wire at a time. That is why its measured in bits. You are measuring physically how much data can transfer.
The packets being sent may or may not confine to the 8bits = bytes. Some mainframes for example had words of 1024bits 0.o
Probably should have opened up the power supply for the hub and blown that out too.. that's probably where a lot of the smell is coming from.
didn't open it but blew plenty of air in there to get most of the dust out!
bro has infinite storage space for old server tech
for now
@clabretro I would kill to live in your closet.
I'd clean and do your heavy works for free just for you to let me watch and play with these machines
A stack nice this time the video wasnt posted at 1am my timezone
Or youtube not recommending it to me.
Awesome!
Bytes weren't always 8 bits, so the link speed using bits is likely because byte sizes couldn't change it. Like how serial uses baud rates.
If you want more of these, look up the Fore/Marconi ES-2810. Exact same switch, just rebranded...
interesting! had no idea
The thing about network speeds dates back to modems, way before the internet. The speed of the modems was always measured baud or bits per second, so it’s kind of a legacy thing which stuck.
Intel (legacy), internet archive and manuals library has interesting material on these switches and hubs
The stack must flow!
The fiber is easy, you see "SX" so that is the standard. First match the two ends to be both the 1000Base-SX type, and then you check whether one hole is emitting light (the camera might be more sensitive to pickup the infrared, but I usually saw a red glow in every system). Then you attach the cable on one end - and re-check the glow on the other end of the cable prior putting it into the device. You obviosly put the light into the dim hole and the dim fiber to the lit hole. Easy :)
One positive of ST connectors at you bulkhead is that it was easy to swap if you didn't get a link. LC and SC require prying apart or more often breaking the end of the cable to swap them.
It looks like a 90s midrange hifi 👍
byte is just a word for 8bits as a convinience. Just as you have 4 bits = nibble, 8 bits = byte, 16bits = word, 32 bits = double word, 64 bits = quad word (which is what most computers run on today)
and some fancy extension can increase this even more for stuff like graphics.
And the actual transfer frames may not even be 8 bit aligned.
If you are looking for a conspiracy you should look into how hard drive manufacturers list their storage space in 10 base instead. So there is where you get things like Mebibytes where you really mean base2 units.
I could not find any wire protocol info for those stack interface modules but it would be quite probable that they are not Ethernet of any kind.
I bet it's just the MII (media-independent interface) plus serial.
Got me wanting to get some of these myself for no reason.
The first cables/connectors... look like Ultra SCSI wide like, meaning 68 pin connector full parallel, hence the bulk of the cable.
I wish you were looking at some 90's IBM gear, that would cause someone other than just me to be trying to find the lost to time Nways Management software...
hopefully someday!
You gotta get into fiber more! :D It's not complicated it's super simple it just seems scary at first but you'll have an absolute blast with all the transceivers you'll be able to play with bro!! You gotta get yourself a job as a Data Center tech on the side, your jaw will drop, it's like being a kid in a candy store for nerds, lol.
Seems like a pretty big security oversight if you can just send a factory reset without a password! haha
right? haha
It uses bit (and nt byte) becasue hsi is the raw transfer speed. The hub device does not care at all about the network protocol used. It simply replicates the eletrical signals to the opther ports. The MAC layer protocol can contain any network protocl inside it. The switch only consider the first seqeucne of bits to form a MAC address, disregardgin anything that comes after it. i.e., just broadcast the electrical signals to the assocaited port. Since every upper layer protocl adds an overhead data, the real transfer speed at the user level does not match the raw speed at the network device. This is why you consider the raw device speed (10/100/1000/10000) which can use any sort of protocls. Also, it has to consdie the signaling protocol. For ecxample, how the device will identify a seqeunceof 10 '0' (zeros)? The bits are encoded on a way that sequences of zeros and ones can be read by the network devices.
Also manuallib has a small grouping of information on these hubs and switches
clabretro: The name of the game is stacking
these devices were for closets not data-centers (notice they are "stackable" and not mountable), hence the bulky cables. also not sure if you were completely joking about the bits thing, but bits are used because not all data sent over a network is part of the data. remember there are multiple network transports/protocols and the overheard in bits of a packet can vary so the division method used is for theoritcal capacity not for actual data rates. nothing to do with marketing at all.
The modern Cisco Stacking Garden-Hoses are about the same in regards to stiffness.
And with 3750X, 3850 and 9300 switches you have the thick powerstack cables as well.
I would put that 1gig slot in the system with the multiplexer for better data split unless it also complains about having the link port. There might be a way to disable that port in software if its not a hardware conflict.
Hexa STACKED. Love it!
These intel express things remind me that I was just browsing through ebay (as one does) and discovered at one point intel did routers too. Only one I saw was the intel express 9250, which is an isdn router, would've been kinda cool to pick up out of curiosity, but $500 is not "curiosity" money :(