I think what you have explained works for RAID 3 & 4 (theoritically , i.e., as they are not used in the industry level ) .But in case of RAID 5 - there will be no DEDICATED PARITY DISK , the parity will be distributed among all the disks used in the RAID .
I have seen a lot of videos and read different explanations available on web. Never was able to understand parity and the reconstruction of Data.. Its all clear now.. And I thank you so much. The best way to explain both Parity and how RAID works..
Carlos, thank you so much for your brilliant explanation! It is definetely worth 12 minutes for those, who want to know about data parity principle. God bless you for sharing your knowledge in such simple way!
oh good Lord! thank you for your explanation!!! i don't know how many tabs of TH-cam video, website reading i had open opened just to get a basic explanation of raid 5. thank you sir.
In the context of a RAID array, I had not found anything explaining the concept of "parity" up to this point. Your video not only cleared up that mystery for me, but now...I also have a relatively firm grasp of how it works. Thanks so much for taking the time to share this!!!
Great refreshment video, thank you! I was just rebuilding the RAID in one of our servers and did not remember how it's done exactly. You just learn so much thing in IT all dey every day.
i saw lot of videos on this topic but first time i saw a clear explaining video on this topic .what a great explaination ! .............thankyou very very much
That's the best explanation of XAND ( Exclusive-AND ), very simple and easy as pie to follow. Reminds me of the days where I learned to program a PLC for two years.
I worked on a 16 bay raid 5 when it was new. It allowed 3 dead drives. Oddly I find this was NOT A common norm and you guys are correct. You can not lose more then 1 without an issue.. thanks guys for pointing it out
You can actually manually get some data back by JUST having a healthy parity drive but it is a VERY long and tedious task. I don't know how to do it but I worked with an admin that was able to get some customer data back over the course of multiple nights.
Thanks a ton. Other videos explained the rebuild process along the lines of "a magical fairy comes and spreads healing dust on your computer". You actually got to the science.
Thanks a million, SIR... I got the Grand Fairy and magical pixie dust rebuilding explanation as well from other sources. I looked for an hour on this (MAGICAL PARITY fixer). I got the understanding of Parity from the ECC SDRAM on the DIMM modules in the slots, and it looks like the same line of logic applies here to the RAID configuration on hard drives as well. Thanks for the Graphical explanation, now I can regain a few hours of sleep. If we ever cross paths, Dinner is on me.
What really happens is of course way more complicated but this is a simple explanation of concept. Glad it HELP!!! I live in South Florida, I'll take that dinner... THANKS
@@biigsmokee But the access time is still just as slow as one drive. It still needs to spin up the discs before it can access the data, but once they are spinning they can take data from multiple drives which increases the speed of reading, but writing is as still as slow as one drive. RAID 0 increase both write and read speed, but they still doesn't increase access time as they still need to spin up the discs first. SSD will still access data faster because it doesn't have to spin up any discs.
For any newbies, the explanation given in this video is only a basic explanation. First of all, digital bytes (8 bits) are read just like decimal numbers, from right to left not right to left. Bytes are base-2 as opposed to base-10 (decimal). Therefore, the bits 00 = 0, 01 = 1 and 10 = 2. The byte given as an example in this video is the decimal number 228. The other key point is that with RAID 5, the data and the parity bytes are written across all available drives. This example tends to imply that the 1/2 of the byte is always written to DISK 1, the other 1/2 of the byte is written to DISK 2, and the entire parity byte is written to DISK 3. But what if you have a 6 disc array? The parity data doesn't just get written to the last drive in the array or every third drive. In reality, the underlying data and the parity data are written across ALL available drives, so the process of rebuilding the array has to look across all surviving disks to rebuild the failed drive. The controller has to rebuild the matrix of underlying data and the parity data.
Parity calculate with XOR "en.wikipedia.org/wiki/Parity_drive". You use XNOR for calculate. But thank you, i understand parity with your video. Regards :)
Your video explanation was good, until the XOR calculation point... When you have two similar values like 1 and 1 or 0 and 0, the output is 0. When you have not matching values, like 1 and 0 or 0 and 1, the output is 1. In your video, you had it reversed...
But my explanation is only a concept of operation. It's not exactly how it works. I could do the XOR but then I have to do the grouping, the sector, the data split when you have 4 or more drives. But thanks for the feedback
How would a RAID-5 work when you have 4 drives? Lets' take the first bit for example DRIVE-1: 1, DRIVE-2: 0 DRIVE-3: 1. So the three drives don't match. In this case, how would parity know how to reconstruct the data? "0" means that the three drives don't match but how would you know if to reconstruct with 100, 101, 110, 011, 010, 001, etc... there are many options for the mismatched bits.
It is a very simplified explanation. There is alot of math and tables stored on each disk. But the basic BASIC way of how raid 5 work IS this way. Parts of data on different disks and a parity bit on disk for error corrections and data rebuilding.
I have read some systems do but using xor would invert the example. But reality us that is only a very simplified example. Raid is way more complex to track it's changes and regeneration status, ect.
@@cashe18 I am just doing a class on storage now, and the teacher is having us calculate the parity of RAID 5 with XOR. Your video was good, but confused me so I had to double check.
Thank you for your question. To make this simple we will only be talking about 2 bits of data. Let's call them A & B. And let's call the parity bit X. So let's imagine data A and B are 0. Parity bit will ask if the 2 bits dont match. Since they do match then parity / X = 0. So A=0 B=0 X=0. If I loss Data A then I can rebuilt it because B=0 and the X answer tells me the other data bit MATCHED. so B data MATCHED A DATA. So the missing answer is 0.... if B=1 and the parity/X=0 which means the data matched then A would have equalled 1. If you parity/X bit=1 then that means that the data bits DO NOT MATCH.. So if A=1 and you are missing B then B would be 0 because our parity told us that the bits did not match. NOW let's say it is the parity bit that got damaged, then we still have our data in place, just the protection is gone. So in rebuilt, the parity is rewritten based on the data bits if they don't match (1) or match (0). This is a very basic explanation where what is happening is way more complicated. Bit A goes to drive 1, Bit B goes to drive 2, and Bit X goes to drive 3. If 1 drive fails, we can rebuilt that missing part based on what the other 2 drives state. I hope this helped
@@cashe18 thanks for the response! This is some pretty complicated stuff, it's amazing what humans have been able to accomplish. That's why smart people make the big bucks!
Well, the video is a very simplified idea of how it works BUT the data is constructed via the raid card. BUT in actual working state, if the unit is in degraded mode, distribute is actually limited. hence the slow performance in degraded status.
Thanks for this it has helped a lot I was really struggling to understand it. So when it goes to write the next byte would it then record the parity data on drive 1 instead so drive 2 and drive 3 become the main drives?
Are you sure there's a speed penalty with RAID 1 versus a single disk? I'm pretty sure there isn't. Even if there is, it's certainly nowhere near as extreme as you claim in the video. And with the right controls, reads are actually faster with RAID 1 than with a single disk.
This is GREAT question. RAID 5 typically was on server board which used HARDWARE Raid cards. These cards like the adaptec brand would actually communicate to all the drives at the same time. leaving the Hardware card as the center point of data in and out of the motherboard. and it's based on the throughput of the drive. As an example imagine each drive can transfer 1MB of data a second. Card reads 4 drives for 1 sec for a total of 4MB of data accessed - 1/3 which is parity error check data. That is 2.66MB of usable data in 1 sec where a mirror would MAX at 1MB in the same 1 sec. Here is the other catch, since a mirror setup is very simple. Many people now a days most doesn't use RAID hardware cards to do mirrors. doing what is called a Soft raid mirror. Motherboard can usually only access 1 drive channel at a time. So reading and writing to 2 drives takes double the work. 50% overhead. When using VERY fast matching drives, it's hard to notices. So even with a hardware RAID set to a mirror (Raid1) the most you can access in 1 sec would be 1MB of usable data. This is why the Unsafe HARDWARE Raid0 striped is the FASTEST. Say you have 4 drives on Raid0 and you access them for 1 sec, you pull 4MB of usable data. These are example numbers but this is the basic theory. But at the end the Computer is SO WAY faster then the data is accessible from a hard drive that it comes down to the quality of the hardware.
Wrong Calculation of Parity = XOR value of the Data. 0 XOR 0 -> 0 0 XOR 1 -> 1 1 XOR 0 -> 1 1 XOR 1 -> 0 And You cannot fail more than 1 Disk in RAID 5.
at 2:35 you said that if you have more drives in raid 5 lets say 6 drives. You said that more drives could fail and all the data could be there. In raid 5 it doesn't matter how many drives you have. If 2 or more fails. You've lost data.
correction.. In Raid 5, the more drives you have , more more failures you can have. in a 3 drive setup, you can lose 1. 4 drives you can lose 1 whole drive and have error on another. 5 disk setup, you can have 2 totally fail on you. Many years ago I did a 12 drive setup. 1 was setup as an active Hotswap. This setup allowed us to have 4 total drive failures and the data was sound. I should have been more clear in that part that depending the size of the raid setup will also determine the allowed failure size. Thank you for your feedback.
Proven here www.computerweekly.com/answer/RAID-5-recovery-What-is-the-maximum-number-of-physical-drives-in-a-RAID-5-configuration You can only lose 1 drive in raid 5 no matter how many drives you have
I'm very surprised they would say this. When I was at viacom, goodness the size of those raid setups and when we would have to do a recovery from multi drive damage would take ALL NIGHT! the drives were SCSI which at the time maxed at 300gb each. Our raid cards were all adaptecs. The exchange servers OS would be on 2 drives setup as a mirror RAID 1, and the data/storages were all large raid-5 setups. Our digital department Raid Racks has status lights for each drive. I can tell you that I myself did so long multidrive fails because our data was SO huge that we had Tape backups that would be roboticly switched. I will look into this, if its because it's IDE but thanks for bringing this to my attention.
There is no “parity drive.” Parity is odd or even and rotates across drives. When you lose a drive you lose 1/n parity bits, where n is the total number of drives.
It's RAID 5 explained with only one byte of data. If he was to write another byte after that, he would've placed the parity part in drive 1 or 2, but since this video is only to show how RAID 5 recovers it's data, there was no need to basically show the exact same thing again with another drive.
I think what you have explained works for RAID 3 & 4 (theoritically , i.e., as they are not used in the industry level ) .But in case of RAID 5 - there will be no DEDICATED PARITY DISK , the parity will be distributed among all the disks used in the RAID .
I have seen a lot of videos and read different explanations available on web. Never was able to understand parity and the reconstruction of Data.. Its all clear now.. And I thank you so much. The best way to explain both Parity and how RAID works..
Carlos, thank you so much for your brilliant explanation! It is definetely worth 12 minutes for those, who want to know about data parity principle. God bless you for sharing your knowledge in such simple way!
Thank you so much. This is the clearest explanation I've found. Simple and to the point.
+777evd That was what I was going for. Thank you for watching
oh good Lord! thank you for your explanation!!! i don't know how many tabs of TH-cam video, website reading i had open opened just to get a basic explanation of raid 5. thank you sir.
In the context of a RAID array, I had not found anything explaining the concept of "parity" up to this point. Your video not only cleared up that mystery for me, but now...I also have a relatively firm grasp of how it works. Thanks so much for taking the time to share this!!!
Love this. Not only explains how it works, but why it works.
That was an AMAZING video. I've not seen a simpler explanation. Thank you for sharing. You're GREAT!
I'm glad it help......
Great refreshment video, thank you! I was just rebuilding the RAID in one of our servers and did not remember how it's done exactly. You just learn so much thing in IT all dey every day.
Thanks for providing such a simple explanation of parity.
Many devices use this concept for error correction..
i saw lot of videos on this topic but first time i saw a clear explaining video on this topic .what a great explaination ! .............thankyou very very much
Thanks for explaining in such a simple way.
That's the best explanation of XAND ( Exclusive-AND ), very simple and easy as pie to follow. Reminds me of the days where I learned to program a PLC for two years.
Excellent !!!!!!! clearest explanation even 2 years later!!!
Great explanation. First person to explain it so I could understand it
very easy way of explanation to understand even for beginners
Thank you for watching
Very well described in a simple way...Thank you
Excellent explanation, you have presented it with the most simplistic approach . Appreciate it. Thanks !!
Best RAID 5 explanation EVER!! Thank you!
Thank you very much sir, this was very helpful.
You can NEVER lose more than one drive and not lose data on RAID 5
I caught that too!
True, it is Raid 6 where the concept is same as Raid 5 but with 2-bit parity.
I worked on a 16 bay raid 5 when it was new. It allowed 3 dead drives. Oddly I find this was NOT A common norm and you guys are correct. You can not lose more then 1 without an issue.. thanks guys for pointing it out
You can actually manually get some data back by JUST having a healthy parity drive but it is a VERY long and tedious task. I don't know how to do it but I worked with an admin that was able to get some customer data back over the course of multiple nights.
How is that possible? More than one HDD failure has occurred on the server.
Thanks a ton. Other videos explained the rebuild process along the lines of "a magical fairy comes and spreads healing dust on your computer". You actually got to the science.
I try to make the complicated into simple English. This is the Basic of RAID-5. Which is why 1/3 of your space is used for parity storage.
Clear and concise. That helped a lot. Thank you.
+Imran Ahmed Khan I tried to explain it in the easiest fashion possible. Thank you for watching.
Thanks a million, SIR... I got the Grand Fairy and magical pixie dust rebuilding explanation as well from other sources. I looked for an hour on this (MAGICAL PARITY fixer). I got the understanding of Parity from the ECC SDRAM on the DIMM modules in the slots, and it looks like the same line of logic applies here to the RAID configuration on hard drives as well. Thanks for the Graphical explanation, now I can regain a few hours of sleep. If we ever cross paths, Dinner is on me.
What really happens is of course way more complicated but this is a simple explanation of concept. Glad it HELP!!! I live in South Florida, I'll take that dinner... THANKS
Very nice & simple explanation. thanks
Super duper explanation. Thank you.
I believe RAID 1 with a hardware controller writes to both drives simultaneously and does not increase access times over non-RAID.
Yeah I was thinking that.. and it can read from both so 2x faster read
@@biigsmokee But the access time is still just as slow as one drive. It still needs to spin up the discs before it can access the data, but once they are spinning they can take data from multiple drives which increases the speed of reading, but writing is as still as slow as one drive. RAID 0 increase both write and read speed, but they still doesn't increase access time as they still need to spin up the discs first. SSD will still access data faster because it doesn't have to spin up any discs.
Thank you, you have explained in very simple way.
Thanks sir, it is very helpful for me.
Glad to hear that
Thank you for explaining this to me. It really helped me understand how RAID 5 works. :-D
Glad it helped.. Thank you for watching
love from India.. thank you very much
Best parity video
For any newbies, the explanation given in this video is only a basic explanation. First of all, digital bytes (8 bits) are read just like decimal numbers, from right to left not right to left. Bytes are base-2 as opposed to base-10 (decimal). Therefore, the bits 00 = 0, 01 = 1 and 10 = 2. The byte given as an example in this video is the decimal number 228. The other key point is that with RAID 5, the data and the parity bytes are written across all available drives. This example tends to imply that the 1/2 of the byte is always written to DISK 1, the other 1/2 of the byte is written to DISK 2, and the entire parity byte is written to DISK 3. But what if you have a 6 disc array? The parity data doesn't just get written to the last drive in the array or every third drive. In reality, the underlying data and the parity data are written across ALL available drives, so the process of rebuilding the array has to look across all surviving disks to rebuild the failed drive. The controller has to rebuild the matrix of underlying data and the parity data.
Loved this video...... Concept is clear now👍
Explained very good. Thank's a lot.
Thank you. it is very easy to understand
Parity calculate with XOR "en.wikipedia.org/wiki/Parity_drive". You use XNOR for calculate. But thank you, i understand parity with your video. Regards :)
thank you so much...
Thank you very much!
Great video man.
Thank you.. I hope it helped.
It's well explained.
Your video explanation was good, until the XOR calculation point... When you have two similar values like 1 and 1 or 0 and 0, the output is 0. When you have not matching values, like 1 and 0 or 0 and 1, the output is 1. In your video, you had it reversed...
But my explanation is only a concept of operation. It's not exactly how it works. I could do the XOR but then I have to do the grouping, the sector, the data split when you have 4 or more drives. But thanks for the feedback
Best explanation in youtube xD Thanks
THX bro you truly helped me
awesome explanation make video on other rsid likee 6 ,10 etc
Holy f ing s*** sir you are gold! Thanks
That's great, but how does that work with 4 disks? How can I store information from 3 drives on one disk?
What I showed was a really REALLY basic explanation. But RAID-5 spreads the data and parity bit all over the place.
How would a RAID-5 work when you have 4 drives? Lets' take the first bit for example DRIVE-1: 1, DRIVE-2: 0 DRIVE-3: 1. So the three drives don't match. In this case, how would parity know how to reconstruct the data? "0" means that the three drives don't match but how would you know if to reconstruct with 100, 101, 110, 011, 010, 001, etc... there are many options for the mismatched bits.
Well, this example is VERY simplified. But reality is that the data is split among all the drives.
Is is this a simplification or really how it works ?
It is a very simplified explanation. There is alot of math and tables stored on each disk. But the basic BASIC way of how raid 5 work IS this way. Parts of data on different disks and a parity bit on disk for error corrections and data rebuilding.
Doesn't RAID 5 use XOR to calculate parity??
I have read some systems do but using xor would invert the example. But reality us that is only a very simplified example. Raid is way more complex to track it's changes and regeneration status, ect.
@@cashe18 I am just doing a class on storage now, and the teacher is having us calculate the parity of RAID 5 with XOR. Your video was good, but confused me so I had to double check.
hi, thnx for the video, but i want to know how it works if we have mere than 3 hard drives pllllllllllz i need the expplaination soon i you can
Thats so genius and simple ...
thank you
How does it know whether the matching numbers are two 0's matching or two 1's matching? Did I miss something here?
Thank you for your question. To make this simple we will only be talking about 2 bits of data. Let's call them A & B. And let's call the parity bit X. So let's imagine data A and B are 0. Parity bit will ask if the 2 bits dont match. Since they do match then parity / X = 0. So A=0 B=0 X=0. If I loss Data A then I can rebuilt it because B=0 and the X answer tells me the other data bit MATCHED. so B data MATCHED A DATA. So the missing answer is 0....
if B=1 and the parity/X=0 which means the data matched then A would have equalled 1.
If you parity/X bit=1 then that means that the data bits DO NOT MATCH.. So if A=1 and you are missing B then B would be 0 because our parity told us that the bits did not match.
NOW let's say it is the parity bit that got damaged, then we still have our data in place, just the protection is gone. So in rebuilt, the parity is rewritten based on the data bits if they don't match (1) or match (0).
This is a very basic explanation where what is happening is way more complicated. Bit A goes to drive 1, Bit B goes to drive 2, and Bit X goes to drive 3. If 1 drive fails, we can rebuilt that missing part based on what the other 2 drives state. I hope this helped
@@cashe18 thanks for the response! This is some pretty complicated stuff, it's amazing what humans have been able to accomplish. That's why smart people make the big bucks!
Can u please explain how data is going to distribute in disk when the raid5 is in degraded State?
Well, the video is a very simplified idea of how it works BUT the data is constructed via the raid card. BUT in actual working state, if the unit is in degraded mode, distribute is actually limited. hence the slow performance in degraded status.
Good Explanation
if suppose i'm using Raid 5 with 3 hard disk, now my one of my hard disk is full how to replace new one....
Molte grazie amico, ora ho capito tutto!
Don't know what you said but thanks for watching.
thanks u r awesome, aditya masurkar.!
Thank you for watching. hope it helped.
Awesome explanation. My 3rd drive failed, but Windows refused to launch. Why is that?
Probably a reading error.
Thanks for this it has helped a lot I was really struggling to understand it.
So when it goes to write the next byte would it then record the parity data on drive 1 instead so drive 2 and drive 3 become the main drives?
This is just a really simple explanation of the function but the data is actually spread all over.
Wonderful!
wrong, parity would be stored as exclusive or, so negation of what you calculated because it's faster
If you want to be more detail, yes. But this also just a layman's example. We know it's way more technical then this. Thanks for your feedback
Excellent
Nice. Thanks
Thanks for watching
thank you sir
Are you sure there's a speed penalty with RAID 1 versus a single disk? I'm pretty sure there isn't. Even if there is, it's certainly nowhere near as extreme as you claim in the video. And with the right controls, reads are actually faster with RAID 1 than with a single disk.
This is GREAT question. RAID 5 typically was on server board which used HARDWARE Raid cards. These cards like the adaptec brand would actually communicate to all the drives at the same time. leaving the Hardware card as the center point of data in and out of the motherboard. and it's based on the throughput of the drive. As an example imagine each drive can transfer 1MB of data a second. Card reads 4 drives for 1 sec for a total of 4MB of data accessed - 1/3 which is parity error check data. That is 2.66MB of usable data in 1 sec where a mirror would MAX at 1MB in the same 1 sec.
Here is the other catch, since a mirror setup is very simple. Many people now a days most doesn't use RAID hardware cards to do mirrors. doing what is called a Soft raid mirror. Motherboard can usually only access 1 drive channel at a time. So reading and writing to 2 drives takes double the work. 50% overhead. When using VERY fast matching drives, it's hard to notices. So even with a hardware RAID set to a mirror (Raid1) the most you can access in 1 sec would be 1MB of usable data. This is why the Unsafe HARDWARE Raid0 striped is the FASTEST. Say you have 4 drives on Raid0 and you access them for 1 sec, you pull 4MB of usable data. These are example numbers but this is the basic theory. But at the end the Computer is SO WAY faster then the data is accessible from a hard drive that it comes down to the quality of the hardware.
Wrong Calculation of Parity = XOR value of the Data.
0 XOR 0 -> 0
0 XOR 1 -> 1
1 XOR 0 -> 1
1 XOR 1 -> 0
And You cannot fail more than 1 Disk in RAID 5.
You're right! The parity in this way would not work with 4 or more disks. I'm amazed by this explanation
Exactly, this video actually confused me
0 - 0 is 0
0 - 1 is 1
1 - 1 is 0 and
1 - 0 is 1
This guy says it otherwise 🤷♂️
Great ideo
at 2:35 you said that if you have more drives in raid 5 lets say 6 drives. You said that more drives could fail and all the data could be there. In raid 5 it doesn't matter how many drives you have. If 2 or more fails. You've lost data.
correction.. In Raid 5, the more drives you have , more more failures you can have. in a 3 drive setup, you can lose 1. 4 drives you can lose 1 whole drive and have error on another. 5 disk setup, you can have 2 totally fail on you. Many years ago I did a 12 drive setup. 1 was setup as an active Hotswap. This setup allowed us to have 4 total drive failures and the data was sound. I should have been more clear in that part that depending the size of the raid setup will also determine the allowed failure size. Thank you for your feedback.
Proven here www.computerweekly.com/answer/RAID-5-recovery-What-is-the-maximum-number-of-physical-drives-in-a-RAID-5-configuration
You can only lose 1 drive in raid 5 no matter how many drives you have
also here serverfault.com/questions/183183/can-i-recover-a-raid-5-array-if-two-drives-have-failed
I'm very surprised they would say this. When I was at viacom, goodness the size of those raid setups and when we would have to do a recovery from multi drive damage would take ALL NIGHT! the drives were SCSI which at the time maxed at 300gb each. Our raid cards were all adaptecs. The exchange servers OS would be on 2 drives setup as a mirror RAID 1, and the data/storages were all large raid-5 setups. Our digital department Raid Racks has status lights for each drive. I can tell you that I myself did so long multidrive fails because our data was SO huge that we had Tape backups that would be roboticly switched. I will look into this, if its because it's IDE but thanks for bringing this to my attention.
There is no “parity drive.” Parity is odd or even and rotates across drives. When you lose a drive you lose 1/n parity bits, where n is the total number of drives.
ur cool :D well explaining
How about software raid?
@M Isa: Software RAID works pretty much the same way, except that all the back-and-forth between the data and its parities takes place in the RAM.
Wrong parity calculation !
XOR Truth table say that :
0+0=0
0+1=1
1+0=1
1+1=0
You are correct. It's on the comments. Thank you for your input and for watching...
1 XOR 1 = 0 , 1 XOR 0 =1
exactly, he had it reversed and confused me for a second lol
Isn't this RAID 4?
It's RAID 5 explained with only one byte of data. If he was to write another byte after that, he would've placed the parity part in drive 1 or 2, but since this video is only to show how RAID 5 recovers it's data, there was no need to basically show the exact same thing again with another drive.