vMotion Improvements!

Here is the situation; you have built out your virtual environment the best way possible. I am talking about the best servers, storage, and network money can buy. From a bench mark perspective it is the best and everything runs perfect. However, you run into a situation where you need to vMotion several servers at a time to shut down a production server for updates.

Then you realize even though you have the best environment you still are limited to a static amount of concurrent tasks. As well, large vm’s with bigger memory configurations would be sluggish or wouldn’t even migrate all. Finally, even though you invested in a huge 10GbE network with your fancy Converged Network it seems there is always a bottleneck. What to do?

Upgrade! The good news is that with vSphere 4.1 the amount of concurrent vMotion task has been significantly increased see chart below. They call it Scalable vMotion. It does several new things:

  • Restructured how vMotion takes advantage of the network.
  • Better optimization of memory handling.
  • Optimized vMotion convergence logic.

The engine allows a throughput of 8GB/sec on a 10GbE link, which is 3 times the performance of version 4.0. This will significantly improve data-center performance and help with larger migrations. As well, this should improve the adoption among cloud service providers and the better acceptance for the private cloud.