Storage /

Aggregation and virtualization: Beyond RAID

By Anne Skamarock
Network World Storage Newsletter, 01/03/01

Over the past few weeks, I have been discussing the benefits of both virtualization of storage and aggregation of storage. Now, I'd like to talk a bit more about how these concepts relate to each other and to an old stand-by, Redundant Array of Inexpensive Disks.

RAID is certainly a form of aggregation. RAID uses a collection of disks of the same type to provide data protection, spreading data across the disks in such a way as to maximize the recoverability of the data if there is a single disk failure. In effect, the RAID controller aggregates the disks and presents a single disk image to host operating systems so that applications never have to know where or how the data are being placed on the storage media.

A RAID subsystem is an example of aggregation combined with virtualization. It presents a virtual volume to the compute host that " looks " like one disk, when in fact the volume is made up of many disks. The virtualization allows for easy interoperability between the host and the storage subsystem.

Networked storage aggregation has two main advantages over RAID: scalability and quality of service (QoS). Until recently, when customers bought a RAID storage subsystem, they usually purchased a great deal more storage than what they needed at the time. Why would anyone spend a huge amount of money for something they were not going to use just yet? Because the costs associated with taking down their main business applications to add the additional storage two or three times a year was far greater than the costs of purchasing lots of storage ahead of time.

Today, companies like XIOtech and Tricord Systems have storage-area network (SAN) and network-attached storage (NAS) products that allow customers to add additional RAID capacity while also including the new capacity in a RAID group, without taking down the business application. We have heard the terms " hot-swap " and " hot-spare " used often with regards to storage systems. Now we have " hot-scaling, " which allows applications continuous access to vital data, while the capacity available to those applications simply grows.

Storage management software companies like Veritas have products that allow customers to grow their file systems, adding capacity on the fly to a SAN, whether it is via RAID or JBOD - just a bunch of disk. The beauty of the software is that it doesn't require all the storage to be of the same class. This allows customers to build aggregations of storage that are appropriate to the business need. While having a common storage management strategy across an enterprise is a good and worthwhile goal, not all applications need the same QoS from the storage.

With business-critical applications, certainly highly functional RAID subsystems are appropriate and may be aggregated and shared among those applications within the business that need a high QoS. The management of the data stored on those subsystems would be similar. But would it make sense for a business to use only highly functional, highly priced storage for all applications within the enterprise? I think not. Within the networked storage infrastructure, you can design in QoS aggregation. If your applications need ease of capacity growth but not redundancy of data, aggregate JBOD on a SAN or purchase less-expensive NAS appliances. QoS is just one reason to use aggregation of storage. You also could use aggregation to address performance, service levels and common management requirements.

The first examples of aggregation certainly came out of RAID, and aggregation continues to be key to scalability advancements for RAID - but don't limit your use of aggregation within your network storage infrastructure only to RAID.