http://www.bladewatch.com/2008/08/26/hp-proliant-dl785-g5-rocks-at-vmware/ Tracking Blades, virtualization and grid Wed, 10 Mar 2010 06:52:54 +0000 http://wordpress.org/?v=2.9.2 hourly 1 http://www.bladewatch.com/2008/08/26/hp-proliant-dl785-g5-rocks-at-vmware/comment-page-1/#comment-44328 Keith Thu, 28 Aug 2008 02:27:28 +0000 http://www.bladewatch.com/?p=3729#comment-44328 really depends on the environment your on and how high you want to stack workloads. i've setup 16 node clusters with a mixture of 2 socket and 4 socket quads (Intel) and would have used 8 sockets if Intel still made them. I started way back with a mix of DL580G2's and DL740's w ESX 2.x and had great success. More procs; more places to schedule VM's and they can get the performance attention they ask for. really depends on the environment your on and how high you want to stack workloads. i've setup 16 node clusters with a mixture of 2 socket and 4 socket quads (Intel) and would have used 8 sockets if Intel still made them. I started way back with a mix of DL580G2's and DL740's w ESX 2.x and had great success. More procs; more places to schedule VM's and they can get the performance attention they ask for.

]]> http://www.bladewatch.com/2008/08/26/hp-proliant-dl785-g5-rocks-at-vmware/comment-page-1/#comment-44327 DGo Wed, 27 Aug 2008 11:26:57 +0000 http://www.bladewatch.com/?p=3729#comment-44327 Don't you need to weigh the options of that consolidation in terms of space, power and support in the fact that you are putting a lot of eggs in one basket. Don't get me wrong, it's a very nice box, but say you are going to run ~96 VM's on it, what about redundancy/HA. As my environment has grown, it has become a big concern for me. Do I get fewer larger boxes(8 socket) and put them in a HA cluster, or spread the load out over more smaller boxes(4 socket). I have a production cluster of three 4socket QuadCore servers with 90 VM's running in it. Monday, one of the servers took a hit while we had a case open and HA failed to work properly. I lost 30 vm's for little over 30 minutes. For production, I can't see the risk, but maybe as a dev/stage environment I can see a larger box like this being useful. Don't you need to weigh the options of that consolidation in terms of space, power and support in the fact that you are putting a lot of eggs in one basket. Don't get me wrong, it's a very nice box, but say you are going to run ~96 VM's on it, what about redundancy/HA. As my environment has grown, it has become a big concern for me. Do I get fewer larger boxes(8 socket) and put them in a HA cluster, or spread the load out over more smaller boxes(4 socket). I have a production cluster of three 4socket QuadCore servers with 90 VM's running in it. Monday, one of the servers took a hit while we had a case open and HA failed to work properly. I lost 30 vm's for little over 30 minutes. For production, I can't see the risk, but maybe as a dev/stage environment I can see a larger box like this being useful.

]]>