Selecting Storage For Server Virtualization - Capacity

In this entry we will return to our series on selecting storage solutions for the virtualized server environment focusing now on dealing with capacity, storing the virtual server images and all their related data. For most environments shared storage is a presumed need, but making sure you purchase a system that can scale to meet future capacity requirements is critical.Most storage managers do a good job of making sure the initial system meets their upfront storage needs. The challenge is the virtual server project tends to be successful and before you know it you have three to four times as many virtual machines as you originally planned for. The reason for this is that in many environments additional virtual machines are "free". There is often a site license for the operating system in place and there are plenty of CPU resources to go around. The one place there is always an "cost" though is storage capacity. Those unplanned virtual machine images need to be stored.

Most shared storage systems that you will purchase for the virtual environment have capabilities to add additional capacity to them. The problem with adding capacity to a traditional storage system is that every time you do so performance gets a little worse. Most storage system's best day is their first day, as you upgrade their capacity they get slower and slower. Each time you add a drive shelf the storage system's compute power has to manage that shelf and those drives in addition to what it was already doing.

The solution is to make sure that the storage for your virtualized environment will meet your performance needs as capacity is added to it. There are three options here. First you could make sure that you always have budget for a second system and be prepared to manage multiple systems. The server virtualization software does provide some tools to help. For example VMware's Storage VMotion will allow you to move live server images from one storage platform to another but you need tools that help you understand what to move, when and where.

The second option is to make sure that the problem never occurs in the first place by making sure that the storage system has the processing power to provide fully rated performance even at 100% drive count. The problem with this approach is that selecting a system of this type may have a higher upfront cost. You have to buy that processing power now. Then it is in place for use as you need it but then it will save you money over time by not having to add more storage systems.

The third option as discussed in our article "What Is Scale Out Storage" uses a scalable storage system that adds storage processing power as capacity is added. These systems may have less initial capacity or storage processing power but then as capacity is added they actually become faster yet still retain a single volume and management structure. In other words their first day is just the beginning. As long as you match your initial performance needs to the day one storage configuration dealing with future growth should not be a problem.

There is another option. Capacity optimization intelligence can be added to the current storage environment to store more data in the same amount of space. We will explore those in the next entry in our series.

Other Entries In the Selecting Storage For Server Virtualization Series

What Storage Is Best For Server Virtualization? Part I

What Storage Is Best For Server Virtualization, Part II

Track us on Twitter: http://twitter.com/storageswiss

Subscribe to our RSS feed.

George Crump is lead analyst of Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. Find Storage Switzerland's disclosure statement here.