shutdown vm's on vmotion

[martijnk]

Hi, I have a question.

Let's say I have two ESX 4.0 servers on one site. The first server is fully loaded with test vm's which are always up and running, the other ESX server has production vm's only and is also fully loaded.

Now, would it be possible when the production server goes down, to vmotion all the machines to the test ESX server and automaticly shutdown the test vm's in order to create enough capacity for the production vm's? Or can this only be done with a 50%/50% load on each server?

Thanks in advance!

[krowczynski]

HI,

have you got HA and DRS configured on your hosts?

MCP, VCP

[martijnk]

Hi,

Thanks for your quick reply. We still need to built everything, but yes HA and DRS are available.

Is DRS all we need to do this?

[krowczynski]

Well DRS will balance the load of you ESX host.

But you can configured in your Cluster settings, per vm.

Go under Cluster / Edit Settings / VMware HA / Virtual Machine Options.

Here you can configure your Host Isolation Response per each vm.

MCP, VCP

[martijnk]

Thanks for your reply.

We are still in the process of building the ESX environment. We need to scale hardware requirements at this point so the question is can we fully load all ESX hosts and still have HA working? I have no experience with DRS.

We plan on having 2 ESX hosts on Site A and 2 on site B, so 4 ESX hosts all fully loaded, 2 production and 2 test/fallback servers (1 on each site).

Now host isolation response will (from what I understand) shutdown the unhealthy ESX host which lost network connectivity. What I want is to shutdown healthy vm's to make place for vmotion of production vm's, isthis possible with DRS?

Thanks again!

[krowczynski]

Well I would suggest not to load each ESX to 100%.

But if you have some ESX host with DRS activated, the hosts with loadbalacing by themselves.

MCP, VCP

[martijnk]

Ok I understand.

But let's say I load both servers to 80%, one ESX hosts crashes and vmotions to the other host, which means I would need 160% capacity to run all vm's.

So with DRS I can specify which vm's have priority above other vm's and give them more recourses.

So we need DRS anyway.

[krowczynski · Accepted solution]

Yes in this case DRS will be very good for you.

**If you find answers were right or helpful, please award points.**

MCP, VCP

[martijnk]

Ok thanks for your help krowczynski!

[krowczynski]

No problem.

Good luck and you're welcome.

MCP, VCP