Comments on: I’ll make that decision thanks! http://blog.nigelpoulton.com/ill-make-that-decision-thanks/#utm_source=feed&utm_medium=feed&utm_campaign=feed with nigel poulton Tue, 20 Dec 2011 19:37:54 +0000 http://wordpress.org/?v=2.8.6 hourly 1 By: Chris M Evans http://blog.nigelpoulton.com/ill-make-that-decision-thanks/comment-page-1/#comment-108 Chris M Evans Tue, 19 Dec 2006 20:59:53 +0000 http://blogs.rupturedmonkey.com/?p=58#comment-108 EMC were well behind on RAID5, they (I believe) are about to do RAID6 on their Symmetrix range, but no plans on the cheap and nasty (sorry, I mean Clariion) storage. Shame, that's where I'd be wanting it - 500 or 750GB "dodgy" SATA drives....Give me HDS and RAID 6 any day. EMC were well behind on RAID5, they (I believe) are about to do RAID6 on their Symmetrix range, but no plans on the cheap and nasty (sorry, I mean Clariion) storage. Shame, that’s where I’d be wanting it – 500 or 750GB “dodgy” SATA drives….Give me HDS and RAID 6 any day.

]]> By: SanGod http://blog.nigelpoulton.com/ill-make-that-decision-thanks/comment-page-1/#comment-107 SanGod Sat, 16 Dec 2006 05:29:14 +0000 http://blogs.rupturedmonkey.com/?p=58#comment-107 EMC has a very bad habit of trying to take the management away where it counts, but the funny part is that I've seen more people do more damage trying to take the management away from EMC than not. There are a number of instances with EMC where it is in fact best to let the Symm place data. IE don't spend too much effort on where the data is on the back-end of the array, if you were paying attention and bought optimizer, it's a moot point anyway. When you're forming metadevices (stripe-sets) it's best to *NOT* let EMC decide where the devices are, as if you tell it to create a 10-way metavolume without specifying which volumes to include, it will scatter them across the array. While you might get a slight (and I mean *VERY* slight) improvement in performance, it doesn't make up for the hellishness of having meta-members scattered all over the array. Again, buy optimizer, it's a small price to pay, and lay out the hypervolumes sequentially, which almost assures you won't have any back-end contention. (As a function of the Enginuity code is to spread wide across the back-end as a default) EMC has a very bad habit of trying to take the management away where it counts, but the funny part is that I’ve seen more people do more damage trying to take the management away from EMC than not.

There are a number of instances with EMC where it is in fact best to let the Symm place data. IE don’t spend too much effort on where the data is on the back-end of the array, if you were paying attention and bought optimizer, it’s a moot point anyway.

When you’re forming metadevices (stripe-sets) it’s best to *NOT* let EMC decide where the devices are, as if you tell it to create a 10-way metavolume without specifying which volumes to include, it will scatter them across the array. While you might get a slight (and I mean *VERY* slight) improvement in performance, it doesn’t make up for the hellishness of having meta-members scattered all over the array.

Again, buy optimizer, it’s a small price to pay, and lay out the hypervolumes sequentially, which almost assures you won’t have any back-end contention. (As a function of the Enginuity code is to spread wide across the back-end as a default)

]]> By: snig http://blog.nigelpoulton.com/ill-make-that-decision-thanks/comment-page-1/#comment-106 snig Thu, 14 Dec 2006 19:55:29 +0000 http://blogs.rupturedmonkey.com/?p=58#comment-106 Well how long did it take EMC² to move away from RAID-S and go to true RAID-5? Many many years... Well how long did it take EMC² to move away from RAID-S and go to true RAID-5? Many many years…

]]>