vSAN Stretched Cluster and caching

[SuparTed]

Dear all, I would like to know if the following statement is correct:

If a data site failure occurs in a vSAN stretched cluster, the cache layer is reduced by 50% due to a 50% reduction is hosts.

Best regards,

Ted

[TheBobkin]

Hello SuparTed​,

"If a data site failure occurs in a vSAN stretched cluster, the cache layer is reduced by 50% due to a 50% reduction is hosts."

If a site becomes completely unavailable due to network partition then not only access to Cache-tier is lost but access to all data on this site + the compute resources at this site will not be able to access the vsanDatastore so the VMs can only run on the remaining site as these nodes would be the only ones to have quorum on the vSAN Objects (as this site has access to the Witness data also). This is why (typically) all data is stored as RAID-1 across sites with a replica of all data on each site in a stretched-cluster.

Bob

[SuparTed]

Hi Bob, thanks for you feedback.

My questions is concerning a situation where a policy PFTT1-SFFT0 is active in a stretched cluster. In case of a complete site failure, the VM's of the failed site will be failed over to the remaining site. I would like to validate, if in this situation the cache layer is reduced by 50% due to a 50% reduction in hosts.

[TheBobkin]

Hello SuparTed​,

PFTT-1-SFFT0 means standard RAID-1 protection across sites e.g. a replica of all data on each site - if you are running all VMs on one site following one site becoming inaccessible then yes they will only have 50% of the clusters total cache resources available.

Bob