The Sansabelt of server storage

In these uncertain times, it makes even more sense than usual to squeeze every last iota of oomph from existing resources rather than buying more. With that in mind, Veritas has announced that the latest version of their CommandCentral software helps to find unused capacity and make it available, while dance partner 3Par is touting the ability of their thin provisioning to combine with Veritas’ software to maximize available space, and even to automatically reclaim unused space in an array.

The partnership between Veritas and 3Par is not unique. The same message can be heard from many vendors, and in general, the strategy has several parts: Find available storage and make it easily visible to the admin, without having to run command line utilities and pipe the results to a spreadsheet; then, help the admin parcel out the newly available storage as needed; and finally, give the admin some level of comfort that critical apps won’t run out of space, now that they’re no longer over-provisioned by a large factor.

Anyone who has experienced a problem on a database server or e-mail server that resulted in log files growing from a few megabytes to many gigabytes in a short period of time can appreciate the need to give a server some extra storage. However, today you can have that extra headroom without devoting a fixed amount of additional space to the server. Instead, thin provisioning can be used to create a large volume that can grow as needed, but is initially only the size of the disk space actually used.

To give a practical example, Windows Server is generally installed in a partition of at least 100GB, and often 200GB or more, especially if Exchange Server or SQL Server is installed. However, the initial install may actually consume only 10GB or so. With thin provisioning, the volume can be set to a virtual size of 200GB or more, but it will actually use only 10GB. As the amount of data stored nears a set percentage of full (often 80 percent), the storage system can either notify the admin to expand the volume, or with more sophisticated systems such as 3Par or Compellent, expand the volume automatically. If a volume is increased to 200GB (for that theoretical log file), once the problem is resolved and the log file deleted, the system can automatically shrink the volume again to reclaim the excess space.

If the admin is willing to trust the system, he no longer has to worry about the capacity of specific volumes. He can buy excess capacity for the array as a whole, which can encompass storage for dozens of servers, rather than having to play it safe for each and every server, only to have 50 to 80 percent of the internal or direct-attached storage on each server go unused.

I know what you’re thinking. While this is fine for companies that already have a SAN in place, what about smaller companies? Well, there’s an answer there too – storage virtualization software such as DataCore’s SANSymphony can provide similar functionality using existing hardware and with a relatively small overall investment.