Stretched Cluster

[elieqt]

Hello,

I have 3 servers in Site 1 and 3 servers in Site 2 connected through 10G L2 with HDD disks for capacity and SSD for caching. Can we configure stretched cluster between sites? can we use RAID1 with stretched cluster or we need RAID5/6?

And if RAID5/6 is mandatory for stretched cluster, should we need all-flash?

Regards,

[TheBobkin]

Hello elieqt​

"I have 3 servers in Site 1 and 3 servers in Site 2 connected through 10G L2 with HDD disks for capacity and SSD for caching. Can we configure stretched cluster between sites?"

Yes, provided the Round-trip latency between the sites is <5ms.

Stretched-clusters are supported for both Hybrid and All-Flash

You will require a Witness running at a 3rd site - there are a few options for this e.g. as a physical box, a VM run on-premises or in the cloud.

"can we use RAID1 with stretched cluster or we need RAID5/6?"

No, not RAID5'd across sites. While it is possible to use RAID5 as the FTM for local protection (SFTT) data is still protected across sites with RAID1 (mirror of data on each site and witness-components stored on the Witness). Using RAID5 as the SFTT FTM would require a) minimum 4-nodes per site and b) All-Flash, so this wouldn't be feasible with your current set-up.

"And if RAID5/6 is mandatory for stretched cluster, should we need all-flash?"

RAID5/6 is not mandatory and is only applied for local protection as per the above, and nope, Hybrid is supported.

Bob

[elieqt]

We don't want to use Raid5 in the stretched cluster, so in this case Raid1 will be a across site (one mirror in Site1 and the second one in Site2). But what if Server1 in Site1 fails? the VM that was running on Server1 will restart from another server in the same Site and read from the second mirror in the second Site? is it correct?

[TheBobkin]

Hello elieqt​,

It will restart the VMs on the same site only if you have VM Host affinity rules set up correctly - always use 'should' rules not 'must' rules as you want this to be able to restart the VMs on the other site in the event of a site/network interlink failure.

Bob

[Jasemccarty]

Check out the Stretched Cluster and 2 Node Guide here: https://storagehub.vmware.com/t/vmware-vsan/vsan-stretched-cluster-2-node-guide​

With 3+3+1 (+1 being your vSAN Witness Host), you can provide the following protection with vSAN 6.6/6.7

Mirroring across sites - PFTT=1

Mirroring within sites - SFTT=1

Failure Tolerance Method = Mirroring

If you only want a single copy of your data in each site, you'd use PFTT=1, SFTT=0, Failure Tolerance Method = Mirroring

I'd recommend checking out the Failure Scenarios section of the linked Guide if you want to be familiar with component placement and failure handling.

[RajeevVCP4]

https://www.vmware.com/content/dam/digitalmarketing/vmware/en/pdf/techpaper/VMware-Virtual-SAN-6.1-S...