This is another recently-published whitepaper written for the folks at Kingston about VMware HA and DRS Capacity Planning.  This paper was yet again a collaborative effort between myself and a great colleague, John Dodge, and started as a pet project while working with a few enterprise customers.  It rapidly evolved out of the obvious misunderstandings around HA, DRS and memory overcommitment.

Memory overcommitment is a great feature of VMware ESX - one that really separates it from the competition (at least for now).  However, I was surprised at the lack of understanding around the limits to which overcommitment was being pushed and the implications when using VMware HA and DRS.  VMware's own somewhat vague messaging is partly to blame, not really getting the information out to customers on reasonable limits and shortcomings.

Anyway, I digress...  This paper delves into the details of properly planning for capacity when leveraging HA and DRS in VI3, and how overcommitment plays into these scenarios.  It also details precisely what HA does and does not do, something I find isn't documented very clearly by VMware. 

The whitepaper can either be downloaded from Kingston's site here, or from my site (below).  Enjoy!

Download:    VMware HA and DRS Capacity Planning Recommendations (2.4 MB)