When leveraging the promise of virtualization it helps to be able to optimize the resources to get the biggest bang for the buck from your existing hardware. How many virtual machines can run effectively on a single server? What steps can you take to improve performance if things get bogged down?
Maximizing VM performance is a worthy goal for any IT administration professional. Finding the right balance between virtual machine density and hardware consolidation while still providing fast server response remains the ultimate sweet spot.
Useful Strategies for Capacity Management
When first setting up a collection of virtual machines on a server, it is tempting to go crazy trying to fit as many VMs as possible on a single box. Doing so can lead to oversubscription and stress hardware resources like RAM, disk storage, and CPU time. While adding more memory and faster processors mitigate this issue somewhat, it is better to right size the number of virtual machines on each server ahead of time.
Spend some time planning for the number of VMs on each server. It is better to start off small when setting up the first virtual machines on a piece of hardware. Get a feel for the hardware utilization rates over time, while allowing the room to grow as needed.
Hardware Requirements for a Virtual Machine
It helps to be able to estimate the hardware requirements for each virtual machine on a server. A good rule of thumb is to allocate the hardware resources typical of an average workstation for each VM. This means up to 250 GB of hard drive space, at least 4G of RAM, and 2 CPUs.
Once you’ve determined how many VMs to run on one server, it is a good idea to add 10 to 20 percent to the amount of RAM and CPU cores to leave some headroom for performance and also allow for growth. Pare down the number of VMs if additional hardware isn’t available. If there is physical space in the server, allocating a separate hard drive for each VM helps the overall performance.
Don’t Forget about Networking Considerations!
While not having enough RAM, CPU power, and hard drives all contribute to poor virtual machine performance, it is also important not to ignore the networking hardware. This can also be a significant source of bottlenecks. Be sure to allocate enough network bandwidth to each server running virtual machines.
Virtual networking is another area starting to take off in IT operations. Intelligent use of virtual networks can be an inexpensive way to decrease demands on your network bandwidth, allowing more VMs to be comfortably hosted on one server.
After deciding to use virtualization in your organization’s data center, it is important to follow these few simple guidelines, insuring performance stays essentially the same as a hardware-only shop. Start small, allow room for managed scalability, and make sure each server has enough network bandwidth to prevent bottlenecks.