Best Practice and Suggestion in combining different VMHost specs into single DRS cluster ?

[AlbertWT]

People,

In my HP Blade c7000 enclosure, have two different Blade Servers:

HP BL 465c G7

AMD Opteron 6172

CPU Cores: 2 sockets x 12 cores = 24 vCPU total

RAM: 96 GB

HP BL 465c G8

AMD Opteron 6378

CPU Cores: 2 sockets x 16 cores = 32 vCPU total

RAM: 256 GB

So how do I combined the two ofthose hardware types into one single DRS cluster for simplicity ?

What's the caveats or the disadvantage when combining those different Blade servers into single DRS cluster ?

Thanks in advance.

[rcporto · Accepted solution]

Take a look here some for benefits: An often forgotten VMware benefit – Pick and Mix | VMware TAM Blog - VMware Blogs

But remember that if you really need protect again hardware failure from one of your host, you must not run more VM than the host with HP BL 465C G7 can support, otherwise some VMs cannot boot in case of failover.

[Alistar]

Hello,

the only thing I can think of is balancing the RAM on your hosts, if the memory modules are compatible with the previous generation - but even that does not matter that much because the DRS can handle the different memory sizes pretty elegantly. Otherwise as Richardson said, keep a soft cap of your VMs to that your weakest CPU can handle, otherwise you could be facing huge ready times on failover (ie oversubscribing the CPU) which is not good.

[AlbertWT]

So how do you soft-capping means ?

at the moment they are now running on its own DRS clustway for each Blade servers.

[Alistar]

If you know the VMs' workload and how they behave I think you can somehow count the physical core:vCPU consolidation ratio (ideally the one that will keep your CPU Ready <5% at all times) and locate that many VMs in your new cluster that would suit the worst-case scenario (ie. all of them running on the less powerful host)