Server virtualization obviously has a lot of advantages, cost cutting, no complex management, and may be its Eco friendliness too. And obviously one of the most important advantage is the ability to create a lot of virtual machines in a speed previously unimagined by the IT fraternity. But anything this good must definitely have certain undesired effects.
Anyway, before we get on to that, lets get the basics right. It doesn’t matter if you are at the beginning of your server virtualization venture or at the end, how many of you actually know what capacity planning for VM means?
- Is it something to do with how much storage you are going to need?
- Or is it about how the existing storage is going to be deployed?
If you are someone who thinks capacity planning is just about demand supply management, then I have to break it to you – ‘Workload = VMs’ is a myth!
Some organizations might get too smart by developing a progressive plan, say- to grow all workloads by 90% over 10 quarters. But trust me it never works out that way.
Capacity planning is a critical process for a virtual environment to maximize the use of the existing hardware infrastructure, plan for future expansion, and prevent the emergence of performance blocks.
But today IT admins start building virtual servers as and when required and end up deploying VMs, resulting in a surplus of unregulated systems which in turn results in a VM sprawl.
This phenomenon of VM sprawl comes from virtual machines that are running and no longer productive. Since VMs are easy to deploy, sometimes it’s common to build a VM and never use it. Or sometimes a VM is used for testing, then forgotten. (Commonly known as Zombie VMs)
Now, in order to identify prospective Zombies, we have to look for VMs running at idle with little or no variation in their load. To do so correctly, however, you need to look at some of the suggestive signs like recent file and resource activity. While these signs can offer hints that point towards a zombie VM, having to continually monitor to see if a VM is no longer used can be quite time consuming.
One last thing to keep in mind with capacity planning are the dreaded choke points. You may have plenty of VM resources available, but if access to them becomes constrained then your resources aren’t doing you much good. However such choke points are more of an VM architectural problem and can be resolved by carefully planning the pathways for your resources are not limited by anyway.
Capacity planning obviously isn’t a one-time event, it is always a ‘still in progress’ kind of a process that committed work so that your virtual environment gains some degree of control. What strategy do you follow to plan your capacity requirements?
