Greetings,
Please note that StarWind VSAN requires dedicated networking for iSCSI and Synchronization and 1 GbE connectivity. See
https://www.starwindsoftware.com/system-requirements and
https://www.starwindsoftware.com/best-p ... practices/ for more details.
How many switches do you have in each location? What you could do is set up StarWind VSAN according to the best practices at Site B and connect the storage to PC3 via iSCSI.
1. Can I include PC 3 within the vSAN network? If not, does StarWind vSAN have any feature that does syncing to remote servers?
I doubt that network requirements between sites are met.
Please see this post
https://forums.starwindsoftware.com/vie ... ind#p30410
NOTE: make sure network redundancy requirements are met between your sites (i.e., 2x switches, 2x independent links between the sites).
2. If I want to have a witness host, will it exceed vSAN Free's maximum node count of 3? If it does, then can I use any one of the PCs acting as a witness host too? Or does it have to be completely separate?
Given the fact that bandwidth requirements are met, you could go with Witness node (see more on Node Majority
https://www.starwindsoftware.com/help/N ... ategy.html). There are several ways towards building the setup with node majority.
1. HA devices replicated between sites, the witness on site A.
+avoid split-brain (
https://www.starwindsoftware.com/blog/w ... o-avoid-it), Site A will be the only synchronized site in case of connection disruption.
-complete storage outage when site A goes down. The majority cannot be formed out of 1 node.
2. HA devices replicated between sites, witness resides on the VM in the cloud.
+no matter what happens to one of the sites, another site can have storage.
-pricey (spinning up an instance in the cloud adds on OpEx), sub 5ms connectivity not only between sites but also to the cloud is a must.
True, 3 nodes is the limit for FREE version, but technically the option 2 is possible with the custom key; you will end up with the GRID-like system (more on GRID
https://www.starwindsoftware.com/grid-architecture-page)). Replication structure could look like this
1. HA1 is replicated between PC1 and 3, its witness is in the cloud or PC1
2. HA2 is replicated between PC1 and 2
a. heartbeat failover strategy can be used if network redundancy requirement is met
b. node majority with witness on PC3 or in the cloud if network requirement within site is not met.
3. HA3 is replicated between PC2 and 3, its witness is in the cloud or PC1
Last but not least. Consider trialing (i.e., set up the POC cluster) if you feel that you want to go with the commercial version.
NOTE: you cannot switch to Free from Trial.