Comments on: Defragmentation, rehydration and deduplication http://www.aboutrestore.com/2009/06/24/defragmentation-rehydration-and-deduplication/ Blogging about backup, recovery and marketing in the storage industry. Fri, 05 Mar 2010 21:32:47 -0500 http://wordpress.org/?v=2.9.2 hourly 1 By: Jay Livens http://www.aboutrestore.com/2009/06/24/defragmentation-rehydration-and-deduplication/comment-page-1/#comment-3355 Jay Livens Thu, 25 Jun 2009 13:27:07 +0000 http://www.aboutrestore.com/?p=614#comment-3355 Thank you for your comments. I agree with you. My point was simply that just because a system "defragments" or "cleans" does not mean that performance will improve. In many cases, these processes are used for housekeeping activities that have nothing to do with improving backup or recovery speeds. The benefit of defrag (or lack thereof) will vary depending on deduplication solution and it is imperative for end users to test this as part of any evaluation. Thank you for your comments. I agree with you. My point was simply that just because a system “defragments” or “cleans” does not mean that performance will improve. In many cases, these processes are used for housekeeping activities that have nothing to do with improving backup or recovery speeds. The benefit of defrag (or lack thereof) will vary depending on deduplication solution and it is imperative for end users to test this as part of any evaluation.

]]> By: W. Curtis Preston http://www.aboutrestore.com/2009/06/24/defragmentation-rehydration-and-deduplication/comment-page-1/#comment-3337 W. Curtis Preston Wed, 24 Jun 2009 23:54:09 +0000 http://www.aboutrestore.com/?p=614#comment-3337 And BTW, the point of my post was not that defragging from any particular vendor fixed anything, but that the core problem is fragmentation, not rehydration. Rehydration is a myth. All a dedupe system is doing is reading a bunch of blocks on disk when it is restoring. They're either getting those blocks from a few places (low fragmentation, mostly disk reads and few disk seeks, good restore speeds), or a lot of places (high fragmentation, lots of disk seeks, bad restore speeds). And the only way you're going to see the difference is to test, test, test. And BTW, the point of my post was not that defragging from any particular vendor fixed anything, but that the core problem is fragmentation, not rehydration. Rehydration is a myth. All a dedupe system is doing is reading a bunch of blocks on disk when it is restoring. They’re either getting those blocks from a few places (low fragmentation, mostly disk reads and few disk seeks, good restore speeds), or a lot of places (high fragmentation, lots of disk seeks, bad restore speeds). And the only way you’re going to see the difference is to test, test, test.

]]> By: W. Curtis Preston http://www.aboutrestore.com/2009/06/24/defragmentation-rehydration-and-deduplication/comment-page-1/#comment-3336 W. Curtis Preston Wed, 24 Jun 2009 23:49:59 +0000 http://www.aboutrestore.com/?p=614#comment-3336 Test everything, believe nothing. And copying 300 GB to and from a dedupe device is a worthless test. They ALL look good when you do that. You have to mimic what you're actually going to do with it in real life both in terms of throughput, capacity, and retention. Anything less isn't testing; it's testing your patience. ;) Test everything, believe nothing.

And copying 300 GB to and from a dedupe device is a worthless test. They ALL look good when you do that. You have to mimic what you’re actually going to do with it in real life both in terms of throughput, capacity, and retention. Anything less isn’t testing; it’s testing your patience. ;)

]]>