Expand existing VSAN cluster

[Milos007]

Hello everyone,

I have a questions about expanding existing VSAN cluster with new server (new harware). 

Let's say we currently have 4 identical HPE Gen 10 servers in a VSAN cluster and that after two years I want to expand that cluster with 1 or 2 new Gen 10+ servers. These new servers will come with different processors, different RAM speeds, as well as different vendor disks (there will probably be disks of the same capacity, which is most important when inserting new servers into an existing VSAN cluster). Is it possible to do an extension with new servers without major problems? What should you pay attention to?

Greeting,
Milos

[ralfklein]

Hi. You should already have EVC mode turned on. Otherwise, you have to turn off all VMs, turn on EVC, and only then can you add more ESXI hosts with other CPUs. We've already done a similar thing with different CPUs on HPE servers. Cluster mit HPE Proliant DL380 GEN10 Gold 61xx mit Gold 62xx erweitert.

 

[Milos007]

Hello,

Thank you very much for your answer. 

Can we create a new cluster under the same center and include the appropriate EVC mode for both types of servers (old and new) and then remove one by one server from the VSAN cluster and one by one add it to the new cluster where EVC mode is enabled?

As for the disks, they must be of the same capacity (caching disks and capacity disks) as in the existing servers and must be the same number in both new servers in order to have equally written vsan objects on all disks in the servers? The disk manufacturer is not important, it is only important that the capacity of the discks is the same?

 

BR,

Milos

[a_p_]

I'm currently not aware that there's vSAN related reason, why EVC cannot be enabled without downtime.

You say that the 4 hosts are identical, so what happens if you select the appropriate EVC mode? Does it show incompatibility issues?
If the new hosts support the same, or more CPUs it should be possible to change the settings to the current CPUs' generation without downtime. One possible reason, why downtime may be required, is if the existing ESXi hosts have "Skylake" CPUs.

André