VPLEX Metro – Server related cost savings

Posted on July 4, 2012 by TheStorageChap in VPLEX

I had some what of an epiphany in a meeting with a large bank of US origin today. We were discussing the ROI of a VPLEX Metro configuration and talked through the usual things in regard to increases in application availability and reductions in operational complexity, but these numbers are not necessarily tangible costs. See previous post on VPLEX – Operational vs. Business Benefits

Whilst discussing this we talked about the server requirements for an active/active vs. active/passive design. At the moment they have for example 8-node clusters in the primary site running the production workload and 8-node clusters in the passive DR site waiting for a DR event to take place. With a VPLEX Metro configuration they could take 4 of the nodes from the Production site and have them clustered with 4-nodes in, the now active, secondary site meaning that they still have  8-nodes performing Production I/O but also enabling them to utilise the storage I/O in the second site and gain 100% storage availability.

In the event of a site disaster they are however now only running on 4-nodes. There are of course some organisations that will insist on having exactly the same performance even when running in a degraded site state for all applications, in which case they may not be able to reduce the server hardware. But the reality is that there are normally a core set of business applications that need to maintain performance and then a lot of other applications that could live with slightly less performance during the disaster.

To mitigate performance impacts they could if required add some additional servers to the cluster in the secondary site or distributed across the sites. Whatever they decide they can still probably reduce the number of servers across the sites by 25-50%.  But it is not only the reduction in server hardware that is the only saving, it is also the proportionate reductions in things like OS and Application licenses, IP and FC port count, power and cooling. When this is multiplied out over 1000’s of physical servers  it can literally add up to millions in Capex and Opex savings.