Comments on: More Discussion on VMware HA Failover Capacity http://blog.scottlowe.org/2007/12/07/more-discussion-on-vmware-ha-failover-capacity/ The weblog of an IT pro specializing in virtualization, storage, and servers Wed, 08 Feb 2012 17:13:47 +0000 hourly 1 http://wordpress.org/?v= By: slowe http://blog.scottlowe.org/2007/12/07/more-discussion-on-vmware-ha-failover-capacity/comment-page-1/#comment-35754 slowe Mon, 25 Feb 2008 16:39:40 +0000 http://blog.scottlowe.org/2007/12/07/more-discussion-on-vmware-ha-failover-capacity/#comment-35754 Skbl, Check out this article: http://blog.scottlowe.org/2008/01/07/vmware-ha-clarification/ Which may help clear things up (or not). Despite all the blogging about it, it still doesn't seem clear exactly how VMware HA calculates failover capacity. Skbl,

Check out this article:

http://blog.scottlowe.org/2008/01/07/vmware-ha-clarification/

Which may help clear things up (or not). Despite all the blogging about it, it still doesn’t seem clear exactly how VMware HA calculates failover capacity.

]]> By: skbl http://blog.scottlowe.org/2007/12/07/more-discussion-on-vmware-ha-failover-capacity/comment-page-1/#comment-35753 skbl Mon, 25 Feb 2008 15:15:53 +0000 http://blog.scottlowe.org/2007/12/07/more-discussion-on-vmware-ha-failover-capacity/#comment-35753 Hi, Interresting article, but I don't understand what's happening with my setup : I have a cluster (1 host failover) build with two identical servers, with 16GB each. As I used DRS for initial placement, there are running VMs on both machines, with memory configurations ranging from 512MB to 2048MB. So, I should also have 16/2=8 available slots, isn't it ? On the first server, there are 8 running machines (and two powered off) and 10 on the second one ( and one powered off). If I try to poweron one more, failover restrictions don't allow me to do so. Strange figures ??? Hi,

Interresting article, but I don’t understand what’s happening with my setup : I have a cluster (1 host failover) build with two identical servers, with 16GB each.

As I used DRS for initial placement, there are running VMs on both machines, with memory configurations ranging from 512MB to 2048MB.

So, I should also have 16/2=8 available slots, isn’t it ?
On the first server, there are 8 running machines (and two powered off) and 10 on the second one ( and one powered off).
If I try to poweron one more, failover restrictions don’t allow me to do so.

Strange figures ???

]]> By: slowe http://blog.scottlowe.org/2007/12/07/more-discussion-on-vmware-ha-failover-capacity/comment-page-1/#comment-34584 slowe Tue, 11 Dec 2007 10:48:32 +0000 http://blog.scottlowe.org/2007/12/07/more-discussion-on-vmware-ha-failover-capacity/#comment-34584 Mike M, I have more than a few customers who want to reclaim hardware and repurpose it running ESX Server in order to reduce their overall hardware costs. In that kind of situation, it's very possible to end up with ESX hosts with differing specs. One key thing about this issue--how VMware HA calculates failover capacity--is that it's not about total RAM utilization. In your situation, if you had two hosts--Host A and Host B--with 16GB of RAM, 16 VMs at 512MB each on Host A, and 4 VMs at 2GB each on Host B, you would be running each host at 8GB RAM utilization. However, VMware HA would calculate failover capacity at 8 slots (16GB / 2GB = 8 slots) and if Host A failed, not all of your VMs would restart on Host B due to "insufficient failover capacity." That's EVEN THOUGH from a maximum RAM utilization perspective you still have enough RAM. I hope this helps clarify why this issue is critical. Mike M,

I have more than a few customers who want to reclaim hardware and repurpose it running ESX Server in order to reduce their overall hardware costs. In that kind of situation, it’s very possible to end up with ESX hosts with differing specs.

One key thing about this issue–how VMware HA calculates failover capacity–is that it’s not about total RAM utilization. In your situation, if you had two hosts–Host A and Host B–with 16GB of RAM, 16 VMs at 512MB each on Host A, and 4 VMs at 2GB each on Host B, you would be running each host at 8GB RAM utilization. However, VMware HA would calculate failover capacity at 8 slots (16GB / 2GB = 8 slots) and if Host A failed, not all of your VMs would restart on Host B due to “insufficient failover capacity.” That’s EVEN THOUGH from a maximum RAM utilization perspective you still have enough RAM.

I hope this helps clarify why this issue is critical.

]]> By: Mike M http://blog.scottlowe.org/2007/12/07/more-discussion-on-vmware-ha-failover-capacity/comment-page-1/#comment-34577 Mike M Tue, 11 Dec 2007 02:00:07 +0000 http://blog.scottlowe.org/2007/12/07/more-discussion-on-vmware-ha-failover-capacity/#comment-34577 Good article. I just dont understand why anyone would build an ESX cluster using hosts of different specs. Maybe I am fortunate but all clusters I have built have been with a minimum of 2 identical hosts. I run balanced VM's on them and each host can handle the full load. If the VM's running on host A consume 8gb, the vm's running on host B will consume the same or less. It would make sense to me that I would need 16gb of RAM if each host was to have the total capacity needed to run all VMs at max RAM utilization. Am I wrong? Mike M http://www.blatbox.com Good article. I just dont understand why anyone would build an ESX cluster using hosts of different specs. Maybe I am fortunate but all clusters I have built have been with a minimum of 2 identical hosts. I run balanced VM’s on them and each host can handle the full load. If the VM’s running on host A consume 8gb, the vm’s running on host B will consume the same or less. It would make sense to me that I would need 16gb of RAM if each host was to have the total capacity needed to run all VMs at max RAM utilization. Am I wrong?

Mike M
http://www.blatbox.com

]]>