When deciding you need to virtualize your network there are some things you need to consider prior to doing so. Trust me, I thought I had every angle covered prior to my first virtualization, and I ran into some issues, so I wanted to give some tips on things to think of that you may not have, prior to making the jump.
- Does my virtualization plan include a single point of failure?
It’s important to plan your virtual server deployment so that the failure of a single host server will not have catastrophic consequences. I’ve seen many companies eager to make money-saving changes, virtualize their entire network onto a single host…or use a single SAN device to store all their information. When designing physical networks it’s easier to see a single point of failure…but sometimes in the virtual world, people overlook this. I worked at a company that has a single SAN device connected to multiple VMware hosts, should a host go down they are fine since they have VMotion, which will transfer everything to another host…yet if they lose their SAN their entire network will come to a grinding halt. So try to eliminate the single point of failure, or it could cost you in a pinch.
- How many guest machines can each host accommodate?
One of the biggest mistakes I’ve seen some administrators make when moving forward on virtualization is overloading the host server. You can not just throw a bunch of virtual machines on a host and assume there won’t be a problem. Though you also don’t need to make sure every physical host has enough for every virtual server, there is some give and take you just need to plan ahead and give some room for growth or high utilization of a server. Every guest machine is going to be different, you’ll need to at least have an idea of where you would like to place each guest machine when you begin the capacity planning process. I used the host specs, the baseline specs of the soon-to-be virtual server, and played a little Tetris-like game, trying to get everything on the server without overloading it and giving it room to grow if needed.
- What is the contingency plan if a host server dies?
While a server failure is never good, its effects are compounded in a virtual environment. I mentioned an above company that had a SAN as a single point of failure, but some hosts can be themselves a single point of failure. I generally recommend that if you have VMware, you purchase VMotion and keep a spare server available. When I did a virtualization upgrade I generally take one of the more powerful servers left over from the virtualization to make it a host. I use the formula…N+1 (N being the number of hosts you need). This way if there is a failover, with Vmotion, the servers will be transferred to another host…it may be slower, but you won’t have downtime…and you’ll have time to fix the problem host.
- What is the most suitable virtualization platform?
Though I have worked with VMware heavily in my background, you really need to evaluate every network differently. There are many different virtualization products out there and each one has strengths and weaknesses, and it’s important to make sure what works for your environment. If not, you may overspend, not get all the features you need, or get underperforming software.
- Are all my applications supported in a virtual environment?
Believe it or not, some applications are not supported on virtual servers. I know that some versions of Exchange Server are supported only on physical servers. Before you begin virtualizing your servers, make sure that your applications will be supported in a virtual environment. I’ve had problems with some software vendors, and they wouldn’t help me because the server was virtualized, so make sure you check prior to making the switch.
Hopefully, these tips along with my other suggestions from [Virtualizing Domain Controllers] will help you with what you need to get started on the virtualization path.