I understand the general concepts of stretched vSANs spent a lot of time reading on tech notes and best practices, etc...
Can't seem to find definitive answers to these questions so hopefully some people in the community with experience can respond positively and help others with similar queries in the future.
1.For All Flash deployments, is a cache absolutely needed? I.e., PCIe Flash cards, or something that's even faster than the SSDs? What are best practices around that?
2. DataSite-A has a 10GbE dark fiber link to DataSite-B and latency is 2ms. Is all the bandwidth needed for vSAN only? Or can we leverage the same 10GBE link for vMotion and VM inter-connectivity?
i) if the same link can be used are they are best practices in throttling or limiting vMotion and VM inter-connectivity ?
ii) if yes any QoS best practices for vSAN bandwidth?
3. Is RAID5/6 preferable over RAID1 due to better storage usage? Or does RAID5/6 suffer 'write penalties' as they do in traditional RAID configurations?
4. In an Always-On SQL setup, node-1 would be at Site-A and node-2 at Site-B. Would the Always-ON traffic share the same 10GbE pipe between Site-A and Site-B?
i) if yes, again , any best practices in throttling or limiting it to not interfere with vSAN bandwidth?
5. If FTT =1, which in Stretched Clusters it has to be, is this applied to all VMs or can some VMs be set to 0 ?
6. Can a witness host (appliance) be physically at DataSite-A or B, with a router/firewall in front of it to 'simulate' as if it is physically somewhere else (ie., it's behind another subnet and reachable only via L3 but physically it's at DateSite-A for example)?
Thank-you
