Single CSV vs. Multiple CSVs

[ted]

I'm planning to add several VMs to my current setup. I have a two node cluster, Starwind VSAN running on Windows 2012 R2. The existing VMs live in a cluster shared volume (CSV). Should I create a new CSV for the new VMs? I've seen some discussion about this in the forums, but the discussions are several years old, and seem to focus on ESXi. I'm not sure I see any clear advantage, but I'm not sure if I know enough to have an opinion. I'd appreciate any thoughts or suggestions. Thanks

[yaroslav (staff)]

Greetings Ted,

You can create a new CSV to avoid the iSCSI protocol being the bottleneck. It makes sense only if you are planning to run multiple VMs on something like an NVMe disk. This is also a way to go if you plan to create a CSV on a new array.
In most cases, for better manageability of the setup (fewer targets, less confusion), expanding a CSV should do it. This is possible if you have some space left.

[ted]

The cluster servers are Dell R730, with 6 TB of storage (all SSD). No NVMe storage. There is an existing CSV already. It sounds like management would be easier with one CSV, and there's no strong reason to add another CSV. When you say "expand a CSV" does that mean using DiskPart and Extend, or is there some other method I should use? Thanks for your help.

[yaroslav (staff)]

First, it is necessary to grow the HA device, then, expand the CSVFS partition. See more at https://knowledgebase.starwindsoftware. ... ty-device/. Script to use ExtendDevice.ps1