pearlyshells
Contributor
Contributor

HA/DRS design

Jump to solution

We have purchased ESX3.5 with DRS/HA for our Production Virtual Environment. Our VirtualCenter 2.5 has 3 Resource Groups (I believe that is what they are called) designed around CPU compatibility and vMotion. The first Resource Group only has 1 ESX3.5 host because it does not have CPU compatibility with any other host. It has 3 DR VMs (using PlateSpin Powerconvert 6.8.2 incremental conversions) working against its respective Production Physical servers. The second has 5 ESX3.5 hosts with over 50 VMs on all the various hosts. The third Resource Group has 2 ESX3.5 hosts and 15 VMs.

What I'd like to do is setup DRS/HA on each of the Resource Groups. With the exception of the first Resource Group (only has 1 ESX Host), I believe I need to cluster the hosts on Resource Group2 and Resource Group3 in its own clusters then activate HA/DRS. Is that correct?

I plan on just staying with the HA and DRS defaults for now.

Just need to be sure that what I am doing is correct and practical. Can anyone advise?

Thanks

0 Kudos
1 Solution

Accepted Solutions
MR-T
Immortal
Immortal

What you're doing is fine.

I've seen plenty of configurations where there's an AMD cluster and an Intel cluster.

Chipset compatability for VMotion/DRS is essential so it's the best way to manage this.

View solution in original post

0 Kudos
8 Replies
MR-T
Immortal
Immortal

You need to create three seperate clusters. Just right click the datacenter object to do this.

Once you've created the 3 clusters, right click them and edit settings to enable DRS & HA

pearlyshells
Contributor
Contributor

so, I need to create a third cluster for the Resource Group that only has 1 host in it?

0 Kudos
Troy_Clavell
Immortal
Immortal

no need to create a HA/DRS cluster for a single host.

pearlyshells
Contributor
Contributor

thanks very much. So, my thoughts to setup HA/DRS is a practical one? Is that what most folks do? What is the most common approach?

0 Kudos
Troy_Clavell
Immortal
Immortal

setting up a HA/DRS pool is the way to go. If gives you failover in the event of a HOST failure, as well as automated load balancing, etc....

0 Kudos
pearlyshells
Contributor
Contributor

Thanks very much. Certainly seemed like the thing to do. Is what I am going to do the configuration that is used often or do others use a different configuraton...that is, group by like Hosts that server specific purposes or some other arrangement? Just curious.

0 Kudos
MR-T
Immortal
Immortal

What you're doing is fine.

I've seen plenty of configurations where there's an AMD cluster and an Intel cluster.

Chipset compatability for VMotion/DRS is essential so it's the best way to manage this.

View solution in original post

0 Kudos
pearlyshells
Contributor
Contributor

many, many thanks. appreciate all your advice

0 Kudos