Comments on: Disks, lots of disks http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/ Technical musings of an entrepreneur. Mon, 03 May 2010 16:57:35 +0000 http://wordpress.org/?v=2.9.2 hourly 1 By: Brian http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/comment-page-1/#comment-370 Brian Thu, 07 Jan 2010 18:17:34 +0000 http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/#comment-370 We are, and it has been rock solid. We've only had one event which resulted in downtime, and it was due to disk failure (3 at once!) than a failure of the unit itself. In fact, it handled it gracefully--instead of letting the data degrade, the unit just stopped all I/O. Pretty clever. The only thing I would do different would have been to make sure we had plenty of hotspares in a pool waiting to go. We originally had it set up with one large RAID-6 array, and by some outrageously rare occurrence, we lost the 3 disks within 3 days before we could get replacements on site. How are you planning running GFS without shared storage (i.e. using internal)? Both servers will need concurrent bus access to the same disks. Using the "EtherSan" (as I call it), is a great way to go. Majorly expandable bandwidth for low low cost. Run with it! We are, and it has been rock solid. We’ve only had one event which resulted in downtime, and it was due to disk failure (3 at once!) than a failure of the unit itself. In fact, it handled it gracefully–instead of letting the data degrade, the unit just stopped all I/O. Pretty clever.

The only thing I would do different would have been to make sure we had plenty of hotspares in a pool waiting to go. We originally had it set up with one large RAID-6 array, and by some outrageously rare occurrence, we lost the 3 disks within 3 days before we could get replacements on site.

How are you planning running GFS without shared storage (i.e. using internal)? Both servers will need concurrent bus access to the same disks.

Using the “EtherSan” (as I call it), is a great way to go. Majorly expandable bandwidth for low low cost. Run with it!

]]> By: Clay Stuckey http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/comment-page-1/#comment-369 Clay Stuckey Wed, 06 Jan 2010 18:01:17 +0000 http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/#comment-369 Looks like a slick little set up you have there. Are you still running that? Do you have any "lessons learned" from it? What would you do differently if you were to redo it today? I was looking at going with 2 discreet servers for GFS and internal storage in lieu of the san and 2 GFS heads. I think I can do about 28TB raid 6 with hot spares, 10GB connectivity and 800MB/s I/O to the network for about $14k. That would give me a bit more redundancy. Looks like a slick little set up you have there. Are you still running that? Do you have any “lessons learned” from it? What would you do differently if you were to redo it today? I was looking at going with 2 discreet servers for GFS and internal storage in lieu of the san and 2 GFS heads. I think I can do about 28TB raid 6 with hot spares, 10GB connectivity and 800MB/s I/O to the network for about $14k. That would give me a bit more redundancy.

]]> By: Greg http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/comment-page-1/#comment-339 Greg Thu, 07 Aug 2008 19:32:03 +0000 http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/#comment-339 RAID-6 is lame. You see, when you get an array with 24 disks in it all at the same time, I will bet you one gigadollars that two of them will fail within a day of each other, thusly rendering your RAID-6 useless. I'll bet you've already had a service interruption. RAID-6 is lame. You see, when you get an array with 24 disks in it all at the same time, I will bet you one gigadollars that two of them will fail within a day of each other, thusly rendering your RAID-6 useless. I’ll bet you’ve already had a service interruption.

]]> By: The Downeys » Blog Archive » VMWare Server on NFS & RedHat Cluster Suite http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/comment-page-1/#comment-235 The Downeys » Blog Archive » VMWare Server on NFS & RedHat Cluster Suite Wed, 08 Aug 2007 18:31:22 +0000 http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/#comment-235 [...] Centralized storage:  EonStor A24F-R2224 [...] [...] Centralized storage:  EonStor A24F-R2224 [...]

]]> By: Piper Christian http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/comment-page-1/#comment-234 Piper Christian Tue, 17 Jul 2007 23:22:50 +0000 http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/#comment-234 Jason sang your post in something that was supposed to resemble Gregorian Chant and then he bowed to it on his knees in front of the computer desk. He says he's humbled by it. Dork. Jason sang your post in something that was supposed to resemble Gregorian Chant and then he bowed to it on his knees in front of the computer desk. He says he’s humbled by it.

Dork.

]]> By: Brian http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/comment-page-1/#comment-233 Brian Tue, 17 Jul 2007 15:28:18 +0000 http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/#comment-233 It's going offsite eventually, it's just easier to set everything up locally and then ship it to the co-lo. It’s going offsite eventually, it’s just easier to set everything up locally and then ship it to the co-lo.

]]> By: OnceUponATime http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/comment-page-1/#comment-232 OnceUponATime Tue, 17 Jul 2007 11:37:39 +0000 http://www.briandowney.net/blog/2007/07/16/disks-lots-of-disks/#comment-232 Hopefully you have an offisite 12 TB too. Dell Hell is coming to an end. All in the name of bits and bytes. Hopefully you have an offisite 12 TB too. Dell Hell is coming to an end. All in the name of bits and bytes.

]]>