This sounds quite a sensible approach - the current approach does depend on all compute hosts being the same spec which is not awesome.
This sounds quite a sensible approach - the current approach does depend on all compute hosts being the same spec which is not awesome.