Hello,
Short answer to a) and b) is unfortunately No.
As a distributed-storage solution vSAN requires a minimum of 3 data-nodes with local storage (or 2 data-nodes + a Witness node in ROBO or Direct-connect implementations).
These nodes create disk-groups from their local storage devices consisting of 1 Cache-tier SSD drive(NVMe, SAS or SATA), and up to 7 Capacity-tier drives (HDDs or SSDs), up to 5 of these disk-groups can be added to each host (their capacity and layout should be consistent on each host in the cluster), these disk-groups' resources are then pooled and create a vsandatastore that can be accessed by all hosts in the cluster.
You can however have a 3-node cluster with disk-groups and then additional hosts in the cluster that have no local storage that can access these storage resources, though this is not ideal nor advisable.
How many VMs with high graphics workloads are you running here?
You might have better results from having passthrough or shared (relatively expensive) GPUs, or local SSDs/NVMe devices as Flash-read-cache (though whether this will benefit largely depends on the workload type):
https://kb.vmware.com/kb/2058983
While NVMe devices are expensive, Enterprise SSDs and HDDS are relatively affordable and vSAN can achieve very high throughput for the price even with relatively small resources so a basic 3-node decked out with these might suffice (again this depends on the workload).
What storage are the current VMs running on and estimated current/required IOPS?
Bob
-o- If you found this comment useful please click the 'Helpful' button and/or select as 'Answer' if you consider it so, please ask follow-up questions if you have any -o
