As midsize businesses continue to grow, they face the challenge of storing more and more data while dealing with cost constraints and having to manage more complex storage environments.

Virtualization is a solution that allows a single resource to run multiple operating systems and multiple applications, reducing the number of server and storage hardware needed—thus cutting hardware costs, lowering power consumption, cooling costs and management costs. Other benefits include: network load balancing, quick migration from one physical server to another and improved scalability.

Array Virtualization

Despite being available for a number of years, users are not as familiar with the advantages of having virtualization capabilities within disk arrays. However, as businesses are realizing the savings that this technology offers—from initial hardware costs to ongoing management costs—the demand for virtualization in disk arrays has increased.

How Virtual Disks Work

In a virtualized array, virtual disks are created from a larger number of disks than in traditional arrays and can have disk groups as large as the number of disks in the physical array. Sliced across the disk group, each virtual disk occupies part of the disks in the group, which can be a different size and is not dependent on the number and size of the disks in the group. Each virtual disk can be any RAID type supported by the array.

Traditional arrays have disk groups set up based on the RAID protection level and the capacity needed for a particular application. The disk group has the same RAID type, and the capacity is based on the number and size of the disks.

Advantages of Virtual Disks

The key advantages of a virtualized array versus a traditional array are the following:

1. Improved performance: In a traditional array, the performance of a disk group, especially for random I/O, is limited by the number of disks in each disk group. If there are multiple disks groups for different applications, for example, there may be times when one disk group is occupied and another is not. 

   In a virtualized array, where the virtual disks for different applications are all in a large group, all the disks in the larger disk group are occupied, even if one application is occupied and another is not. On average, a virtualized array puts more disks to work at any time, resulting in a higher average performance than a traditional array.

2. Virtual sparing: Traditional arrays have hot spares set aside in the array. The disks are unused until needed to replace a failed disk.  When a disk fails, a hot spare is used to replace the failed disk.  The controller, using the RAID parity protection, rebuilds the data from the remaining disks in the group and writes the data to the spare disk.  A virtualized array virtualizes spare space across all the disks in the group, utilizing all disks in the group. When a disk fails the data is calculated and written to the virtual spare across all remaining disks in the group.
3. Easy and efficient expansion: With a traditional array, when more disk space is needed for capacity or performance reasons, another disk group is created and then connected to the previous group. As new data is added, it is written to these new disks that begin to be used. Unfortunately, it could take some time, depending on the growth rate of new data, before enough data is written to the new drives for them to help improve performance.  (For example, if 1TB of new disks are added and data is growing at 10GB per week, it could take a year before the disks are half full.)

   A virtualized array can easily expand disks within the group, if space is available. As new data is added to the virtual disks, the data is written across all the disks in the group. And if more disks need to be added to the group, they can be added in any amount—one, two or more. Once the disks are added to the group the data is automatically spread across all the disks in a background operation, so all the disks are employed. As new data is added, it is spread across all the disks in the group.  The result is that all the disks are employed for overall better performance.

4. Efficient snapshots: When doing local replication, traditional arrays reserve snapshot space equal to the logical unit that the snapshot is being made of. For the life of the snapshot, it is likely that only a relatively small amount of the snapshot space will ever hold data, leaving the rest of the space unused. In a virtualized array, the snapshot only occupies the space that is needed to hold the snapshot data. This is a much more efficient use of disk space.
5. Disk usage efficiency and versatility: Several of the advantages above show how virtual disks improve efficiency. To improve the efficiency of disk usage, virtual disk sizes can be adjusted to the capacity needed, with any RAID type per virtual disk, and do not have to be a multiple of the disk size used. Easy and efficient expansion allows users to avoid over provisioning, as they would with traditional arrays, while snapshots also add to the efficiency. 

All of these reasons add up to more efficient use of the virtual array versus a traditional array. Although some of these tasks may appear to be too complex to implement in small to midsize business environments, the use of management software can easily enable an IT manager with limited storage expertise to take advantage of virtualization technology. Depending on the operating system, the growth and reduction of virtual disks can be easily automated. The user can simply create the rules to grow or shrink and define thresholds which will automatically trigger the operation. In active data centers, where storage requirements change often, these automated operations can save IT managers a considerable amount of time and keep the efficient use of the virtual array very high.

As potential buyers compare the available mid-range arrays they should look closely at virtual arrays, which have traditionally been limited to larger enterprises. Virtual arrays offer some definite advantages over traditional arrays that can help cost-conscious midsize businesses make more efficient use of the storage they purchase—by reducing the need for more storage, thus saving on hardware, power, cooling and management costs.

Jim Gurnick is product marketing manager for the StorageWorks Division of Hewlett-Packard.