For years we have moved to more powerful more expensive processor and SAN based systems in the goal to get to a more reliable system. We cluster machines together, use proprietary time slicing software to perform backups for each HA system. However it seems we are almost at a point where we are going to collapse back to the simple world again (again we probably have to thank Google and Amazon for this!).
I suspect in the near future gone are the days of setting up expensive OS (Windows or otherwise) clusters, and move in dedicated cluster appliances. Each running a set of host ‘machines’ that can ‘transparently’ fail over. It seems that as hard as the various OS vendors have tried, it appears the VM approach seems to have won over the fickle IT crowd.
As we would expect it wasn’t necessarily the technical skills that wows the crowds, but really the $. Enterprise versions of the various OS’s often cost more in licenses and more to maintain. In addition they are more complicated than managing a single machine. So what is the downside of managing a highly available single machine?
Has anyone tried to image and restore a physical clustered machine? It certainly wasnt trivial when some friends of mine tried. VM’s are so nice and easy to deal with – “Need to add a drive? Ok then I’ll allocate one for you”.
Roll on Simplicity is what I say!
Gareth