Hardware Analysis Community Forums http://www.hardwareanalysis.com/content/topic/67628/ 2008-11-22T17:29:20-05:00 http://www.hardwareanalysis.com/content/topic/67628/ http://media.hardwareanalysis.com/halogo.gif text/html 2007-11-16T04:08:11-05:00 brian pope http://www.hardwareanalysis.com/content/topic/67628/#519274 Regarding defraging RAID arrays-<br /> <br /> Mirrored arrays- while the data on your mirrored arrays will be identical the location of the data on each drive won't be. This is true prior to defragmenting &amp; afterwards, mirroring only means the data itself is mirrored not it's location on each hard drive, the MFT provides the OS with the location.<br /> <br /> Stripped arrays- Obviously the data can be anywhere on each hard drive, a file can also be located on just one drive or spread out over numerous hard drives. When defragmenting RAID arrays the defrag software sees the RAID environment just as the file system does, meaning it defragments the virtual drive. Take a basic two-drive array, it would defragment the pieces of the file on each separate drive as though they were separate files &amp; move each piece of that file as directed by the RAID controller. If piece 1 &amp; 3 were on drive A &amp; piece 2 &amp; 4 were on drive B, your defrager would try to put pieces 1 &amp; 3 together on drive A &amp; the same with pieces 2 &amp; 4 on drive B.<br /> <br /> Your thinking they're already together when the file was created but that's not always the case. This takes you right back to why fragmentation occurs &amp; why you should defrag your hard drives. The adding &amp; removing of files creates spaces between files &amp; the next newly created file will take it's place in the first available empty space. text/html 2007-08-29T02:16:17-05:00 CrAsHnBuRnXp http://www.hardwareanalysis.com/content/topic/67628/#507520 Right. The defragger would move the file(s) to their new (same) location on the hard drives so that way they can be accessed when needed. Thats the way I would think it would work. text/html 2007-08-28T21:37:37-05:00 Dragon Peaceful http://www.hardwareanalysis.com/content/topic/67628/#507446 Mark Allen said: <div class="quote">The same probably occurs in defragmentation as it does when data is written to the hard drive. The blocks are sent to an identical location on both hard drives so that they can be retried for later use. If each block on the hard drive was moved to two (or more) separate locations during defragmentation, the data would get corrupt. </div><br /> <br /> You mean it's put in identical location (cylinder) in both hdd, even when the file blocks are moved during defrag?<br /> text/html 2007-08-28T19:01:36-05:00 CrAsHnBuRnXp http://www.hardwareanalysis.com/content/topic/67628/#507415 Dr. Peaceful said: <div class="quote">I have a question. I am wondering what really happen (in block level) when you do defragmentation to a RAID array?<br /> <br /> In a single hard drive setup, defragmentation will re-arrange file blocks, so blocks related to the same file will stay close together, hence improve accessing performance. Defragmenters also does compaction, so unused space can be more accessable.<br /> <br /> Now in a RAID array, say RAID 0, for example. It's striped, with assigned strip size -- e.g. 16k, 32k, 64k, 128k, etc. So files, especially large ones, are split into 2 or more hard drives. Logically, in user's eyes, it appeared as a single drive. But really, when you do defragmentation, it occurs in multiple hard drives! Remember files are split across drives, how the hack it remembers where to find related file fragments in each drive after it's been re-arranged?<br /> <br /> I am a bit confused. Somebody please explain.<br /> </div><br /> The same probably occurs in defragmentation as it does when data is written to the hard drive. The blocks are sent to an identical location on both hard drives so that they can be retrieved for later use. If each block on the hard drive was moved to two (or more) separate locations during defragmentation, the data would get corrupt. text/html 2007-08-28T18:45:53-05:00 Dragon Peaceful http://www.hardwareanalysis.com/content/topic/67628/#507412 I have a question. I am wondering what really happen (in block level) when you do defragmentation to a RAID array?<br /> <br /> In a single hard drive setup, defragmentation will re-arrange file blocks, so blocks related to the same file will stay close together, hence improve accessing performance. Defragmenters also does compaction, so unused space can be more accessable.<br /> <br /> Now in a RAID array, say RAID 0, for example. It's striped, with assigned strip size -- e.g. 16k, 32k, 64k, 128k, etc. So files, especially large ones, are split into 2 or more hard drives. Logically, in user's eyes, it appeared as a single drive. But really, when you do defragmentation, it occurs in multiple hard drives! Remember files are split across drives, how the hack it remembers where to find related file fragments in each drive after it's been re-arranged?<br /> <br /> I am a bit confused. Somebody please explain.<br /> text/html 2007-08-05T23:08:34-05:00 sovietdoc http://www.hardwareanalysis.com/content/topic/67628/#504306 what about other levels of RAID like RAID 10? It's becoming more popular nowadays so maybe you should also include it into your nice guide? text/html 2007-07-26T21:43:51-05:00 Meats_Of_Evil http://www.hardwareanalysis.com/content/topic/67628/#503039 I must thank you Mark for this information. I only knew a little about Raid 0 and that's it, but thanks to you I might Raid my Pc just for the heck of it <img src="http://media.hardwareanalysis.com/smilies/smile5.gif" width="14" height="14" border="0" alt=":P" title=":P"> text/html 2007-07-24T00:27:15-05:00 phil http://www.hardwareanalysis.com/content/topic/67628/#502604 how about adding what stripe width, size, columns, and are and how changing the stripe size will affect read/write/positioning speeds in different arrays being used, and how they are being used (large database, small file copies, etc)<br /> <br /> .. if you'd like some assistance let me know text/html 2007-07-23T17:06:26-05:00 Lou Bot http://www.hardwareanalysis.com/content/topic/67628/#502541 No RAID10? not that it is important but here is more info from wiki website<br /> <br /> Nested RAID levels<br /> Main article: Nested RAID levels<br /> Many storage controllers allow RAID levels to be nested. That is, one RAID can use another as its basic element, instead of using physical drives. It is instructive to think of these arrays as layered on top of each other, with physical drives at the bottom.<br /> <br /> Nested RAIDs are usually signified by joining the numbers indicating the RAID levels into a single number, sometimes with a '+' in between. For example, RAID 10 (or RAID 1+0) conceptually consists of multiple level 1 arrays stored on physical drives with a level 0 array on top, striped over the level 1 arrays. In the case of RAID 0+1, it is most often called RAID 0+1 as opposed to RAID 01 to avoid confusion with RAID 1. However, when the top array is a RAID 0 (such as in RAID 10 and RAID 50), most vendors choose to omit the '+', though RAID 5+0 is more informative.<br /> <br /> <br /> [edit] Common nested RAID levels<br /> RAID 0+1: Striped Set + Mirrored Set (4 disk minimum; Even number of disks) provides fault tolerance and improved performance but increases complexity. The key difference from RAID 1+0 is that RAID 0+1 creates a second striped set to mirror a primary striped set. The array continues to operate with one or more drives failed in the same mirror set, but if two or more drives fail on different sides of the mirroring, the data on the RAID system is lost. <br /> RAID 1+0: Mirrored Set + Striped Set (4 disk minimum; Even number of disks) provides fault tolerance and improved performance but increases complexity. The key difference from RAID 0+1 is that RAID 1+0 creates a striped set from a series of mirrored drives. The array can sustain multiple drive losses as long as no two drives lost comprise a single pair of one mirror. <br /> RAID 5+0: A stripe across distributed parity RAID systems <br /> RAID 5+1: A mirror striped set with distributed parity (some manufacturers label this as RAID 53) <br /> text/html 2007-07-23T15:45:23-05:00 Beavis Khan http://www.hardwareanalysis.com/content/topic/67628/#502537 _MD_ said: <div class="quote">I had a bad experience with RAIDs... I used to have a SCSI RAID setup and one disk failed. So alright, I insert a new disk, but the crash somehow damaged the RAID controller as well. So I was on the market for a SCSI RAID controller... which was a legacy device (has been used by the company for years) and was not in productions anymore (for a long time actually, so couldn't find even used). So <u>all</u> of the data was lost. Since then, we have a mirrored tape (and now NAS) backup at the end of each day...<br /> Now I know there might be comments about this, but I guess the company is just somewhat skeptic to give RAID another try... lol</div><br /> <br /> This is definitely the main pitfall with hardware RAID 5. In fact, if your write throughput requirements aren't too high, I'd generally recommend doing software RAID5 instead, simply because there is no proprietary hardware to fail, and the array can (at least in theory) be used/reconstructed on any computer. Both Linux and Windows server (both 2000 and 2003) include decent tools for managing software-based RAID arrays. You have a lot more flexibility under Linux, but of course, this comes with the price of additional complexity.<br /> <br /> Regardless though, RAID is <i>never</i> (never!) a substitute for a robust backup scheme <img src="http://media.hardwareanalysis.com/smilies/smile1.gif" width="14" height="14" border="0" alt=":)" title=":)"> text/html 2007-07-19T13:34:37-05:00 _MD_ http://www.hardwareanalysis.com/content/topic/67628/#502129 Adam K. said: <div class="quote">They need to make this a sticky in the Hard Drive forum, but not on the Latest Topics, that's too crowded with stickys. </div><br /> True that<br /> <br /> I had a bad experience with RAIDs... I used to have a SCSI RAID setup and one disk failed. So alright, I insert a new disk, but the crash somehow damaged the RAID controller as well. So I was on the market for a SCSI RAID controller... which was a legacy device (has been used by the company for years) and was not in productions anymore (for a long time actually, so couldn't find even used). So <u>all</u> of the data was lost. Since then, we have a mirrored tape (and now NAS) backup at the end of each day...<br /> Now I know there might be comments about this, but I guess the company is just somewhat skeptic to give RAID another try... lol text/html 2007-07-19T06:22:15-05:00 FordGT90Concept http://www.hardwareanalysis.com/content/topic/67628/#502093 RAID0 = Fast but high risk<br /> RAID1 = Slow but low risk<br /> RAID 0+1/10 = Hybrid of RAID0 and RAID1 being about as fast as RAID0 with the redundancy of RAID1 at the expensive of inefficiency (lots of drives required)<br /> RAID5 = Pretty fast with mediocre risk (can lose one drive but no more). RAID5 is XOR mad which, if you don't have a good RAID card, that means a lot more work for the CPU. I would only ever consider RAID5 if you got a good RAID card that can handle it. RAID5 is far more efficient than RAID 0+1/10 in terms of read/write speed and capacity. Should a drive fail, RAID5 takes a long time to rebuild the lost drive.<br /> <br /> If you want mad write performance on RAID5, use Write Through (write requests take immediate priority over read requests). If write performance isn't critical, use Write Back (caches all write requests and writes when it can). text/html 2007-07-09T17:31:45-05:00 CrAsHnBuRnXp http://www.hardwareanalysis.com/content/topic/67628/#500816 Big Beavis said: <div class="quote">RAID-1 can withstand multiple drive failures.<br /> <br /> This is true only if you have more than 2 drives in the array (which would be unusual, but certainly not impossible).<br /> <br /> Regarding RAID 5/6, another significant disadvantage is (relatively) lousy write performance. If your application is a write-intensive database, you'd be much better off going with RAID 1, or a hybrid approach (eg RAID 1+0). <br /> <br /> It might also be worth mentioning that many &quot;real&quot; RAID controllers will allow you to designate a drive as a &quot;hot spare&quot;, meaning that in the event of a member disk failure, the controller will add the hot spare into the array, and begin rebuilding right away. This is usually done only with RAID 5 arrays.<br /> <br /> Edit - one more minor nitpick - the title should be &quot;Arrays&quot;. You don't need an apostrophe.</div><br /> Multiple drive failures meaning that you can only lose one drive multiple times in that array.<br /> <br /> Good catch on the &quot;Arrays&quot;. I didn't even notice that. I'm not sure why I did that in the first place. text/html 2007-07-09T14:55:17-05:00 Beavis Khan http://www.hardwareanalysis.com/content/topic/67628/#500789 Mark Allen said: <div class="quote">RAID-1 can withstand multiple drive failures.</div><br /> This is true only if you have more than 2 drives in the array (which would be unusual, but certainly not impossible).<br /> <br /> Regarding RAID 5/6, another significant disadvantage is (relatively) lousy write performance. If your application is a write-intensive database, you'd be much better off going with RAID 1, or a hybrid approach (eg RAID 1+0). <br /> <br /> It might also be worth mentioning that many &quot;real&quot; RAID controllers will allow you to designate a drive as a &quot;hot spare&quot;, meaning that in the event of a member disk failure, the controller will add the hot spare into the array, and begin rebuilding right away. This is usually done only with RAID 5 arrays.<br /> <br /> Edit - one more minor nitpick - the title should be &quot;Arrays&quot;. You don't need an apostrophe. text/html 2007-07-04T04:02:49-05:00 Adam Kolak http://www.hardwareanalysis.com/content/topic/67628/#499983 Nice write-up.<br /> <br /> They need to make this a sticky in the Hard Drive forum, but not on the Latest Topics, that's too crowded with stickys.