Virtualization of resources bring some interesting issues to the table. Not the least of which, is where the physical locations of your compute resources are at any given moment of the day.
The point of virtualization is that the systems you use are no longer tied to a specific piece of physical hardware, things can move quickly and without notice. For example, a resource located physically next door to you today could be moved via sVmotion to a server across the country tomorrow. As long as the networking team does all the appropriate routing changes, you’d never know.
There are lots of potential issues to consider, but three are:
1 – If you’re servers are not local to you, then the staff responsible for managing those resources at the current time may also not be local. This means that you’ll have to coordinate across time zones to perform maintenance and other tasks.
2 – Flipping resources to another datacenter may mean you suddenly lose physical access to your systems. The good news is that you can always flip the resources back if something goes physically wrong and you don’t have anyone at the other location at that time to plug the wire back in.
3 – Especially for international companies, technologies that cannot be exported could accidentally end up on virtual systems housed in a non-export country. If you deal with encrypted data-sets, this could become a very serious problem.
When you discuss cloud, the situation gets even more confusing, as you may literally not know what physical location your systems reside in at any given time. SLA’s with the cloud provider become absolutely vital, and must be reviewed regularly.
Separating the compute power from physical hardware is – overall – a good thing, but for as many problems as virtualization solves, we do have to remember that there are new problems to consider. Geography is one of those problems.
Dust off your maps…
Photo Credit: Norman B. Leventhal Map Center at the BPL