I am working with a client that is redesigning all their ESXi Hosts. I have posted another thread here: Datacenter ESXi Environment Redesign
That community doesnt get the traffic that this one gets. I was hoping a few would pop over and review it.
Quick version: I am replacing hundreds of old hosts with:
The issue I am having, the old hardware is so bad that I am getting a 10 to 1 return with the new spec. This would put about 2000 VMs per datacenter on 10 New Hosts... am I missing something or is this the new world we live in.... OR is this classic case of to many eggs in one big basket. Right now all my DRS and FT numbers are adding up.. this works. Its basically GO BIG OR GO HOME design.
Thoughts, opinions and reviews are welcomed.
So from the above it seems like you are able to reduce the server count down, for every 10 older ESX hosts you can replace with a single newer host ?
What are you using to produce your consolidation ratio ? A key bit of information missing from the post would be your average VM spec and the Virtual : Physical ratio (vm's per host) seems like 200:1 if looking at your other post example.
Concerns, on the face of it, Such a heavy reduction in hosts could become problematic for Storage HBA throughput, you mention large amounts of data, NIC thorough put, while 10GB is used you are reducing the count drastically and have NFS on this , CPU over subscription, vCPU to pCPU co-stop do to physical CPU reduction. RPO / RTP, while FT is being used you have to think that now a single host failure would mean a interruption of 200 Guests, previously just 20..
But its rather difficult to give advise on data I haven't seen though, with designs this size id suggestion as much review as possible.