Comments on: The coming hard drive capacity wall? http://silvertonconsulting.com/blog/2009/09/21/the-coming-hard-drive-capacity-wall/?utm_source=rss&utm_medium=rss&utm_campaign=the-coming-hard-drive-capacity-wall Storage, Strategy & Systems Thu, 09 Feb 2012 20:45:58 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Thinker http://silvertonconsulting.com/blog/2009/09/21/the-coming-hard-drive-capacity-wall/#comment-22 Thinker Thu, 05 Nov 2009 23:52:37 +0000 http://www.silvertonconsulting.com/blog/?p=714#comment-22 Who uses RAID5/6 with SATA drives to store anything important? I think those types of RAID will go out of fasion - check out the IBM XIV for instance. Thinker, So XIV uses RAID 1 protection. Does that mean that RAID5/6 couldn't be used, I don't think so. In my mind, Raid 1 provides a bit faster rebuild time but it costs you 50% of your available storage to get there. Can it be done with less, I believe so, and the techniques I discuss in my <a href="http://www.silvertonconsulting.com/blog/2009/10/23/are-raids-days-numbered/" rel="nofollow">Are RAID's days numbered post</a> can also help. Ray Who uses RAID5/6 with SATA drives to store anything important? I think those types of RAID will go out of fasion – check out the IBM XIV for instance.

Thinker, So XIV uses RAID 1 protection. Does that mean that RAID5/6 couldn’t be used, I don’t think so. In my mind, Raid 1 provides a bit faster rebuild time but it costs you 50% of your available storage to get there. Can it be done with less, I believe so, and the techniques I discuss in my Are RAID’s days numbered post can also help.
Ray

]]> By: Karl Katzke http://silvertonconsulting.com/blog/2009/09/21/the-coming-hard-drive-capacity-wall/#comment-21 Karl Katzke Mon, 05 Oct 2009 18:33:36 +0000 http://www.silvertonconsulting.com/blog/?p=714#comment-21 Doesn't matter. Who's going to buy a drive that large? Not enterprise customers; 1TB drive RAID rebuild times are already measured in hours, during which you can have another drive failure. You end up scaling out using additional parity stripes and having hot spares so that you can start rebuild right away, but that doesn't scale as drive sizes increase. (Hence the slow adoption of 2TB drives in the enterprise.) It'll be interesting to see what space these large drives occupy in the enterprise. What really needs to be increased isn't read head density but practical read/write and seek speed to get the array rebuild times down. Karl, There are other options to RAID5-6. Some of these can speed up rebuild rates. Let's not forget that RAID was originally intended to help small disks compete with the reliability of large (expensive) disks. More reliability from your disk drives could also minimize the time spent in rebuilding.... Ray Doesn’t matter. Who’s going to buy a drive that large? Not enterprise customers; 1TB drive RAID rebuild times are already measured in hours, during which you can have another drive failure. You end up scaling out using additional parity stripes and having hot spares so that you can start rebuild right away, but that doesn’t scale as drive sizes increase. (Hence the slow adoption of 2TB drives in the enterprise.)

It’ll be interesting to see what space these large drives occupy in the enterprise. What really needs to be increased isn’t read head density but practical read/write and seek speed to get the array rebuild times down.

Karl,
There are other options to RAID5-6. Some of these can speed up rebuild rates. Let’s not forget that RAID was originally intended to help small disks compete with the reliability of large (expensive) disks. More reliability from your disk drives could also minimize the time spent in rebuilding….
Ray

]]>