The number of users is important, but just as important is the amount of IOPs required for each user. Has that sizing been done?
Once the IOP requirements are set, acceptable latency levels need to be established. Depending on the FAS model deemed most fitting for the workload, VST Flash (FlashCache or FlashPool) may suffice. But we can't really give a good answer until the sizing has been done. If you need help with that, let us know!
TME @ NetApp
All-Flash FAS (AFF)