We talked about what server virtualization is and how a properly configured virtualization environment can greatly help keep your workloads/services up and available. There are a dozen ways to configure each vendor’s hypervisor properly, so we’re going to speak in general terms of the technology to examine some of the common issues that we encounter on a daily basis.
The importance of a virtualization host is the sum of its parts. If you run your email, databases, directory structure and web servers all on a single box, that box is as important as all of those programs together. If that server goes down, your business functions are basically dead in the water. Similarly, if you have a whole bunch of servers that on their own aren’t very important, but all of them added together might make the overall host critical because having all of those systems down at the same time is unacceptable. With this in mind, purchase hardware accordingly, and allocate an appropriate amount of time configuring and maintaining it.
Network ports are cheap so there shouldn’t be much excuse to not do the following. Make every networking connection redundant: virtual machine (VM) traffic, management, and clustering traffic (be it for VM migration or whatever your vendor requires). When combining multiple things onto network interface cards (NICs) always think in terms of Denial of Service (DoS). If you are migrating a VM (or multiple VMs) onto a NIC or NIC pair that is also the “service” traffic or “management” traffic, odd/bad things could happen like the ports being flooded with data. Separate and segment where it makes sense. I recommend segments such as: Migration, Management and Data/Services. However, there are other options as well.
At Mirazon, we’re dedicated to efficient and highly available server environments. Contact us to learn more about how we can manage your network and servers for optimal performance.