Match the cost of disk to the value of data. Enterprise data can vary dramatically in its value to the enterprise, so choose disk products that suit the importance of that data. For example, if you intend to store mission-critical database content on the disk array, select high-performance, high-reliability Fibre Channel (FC) hard drives for the array. For less critical documents or other corporate media, you may opt for SAS or high-volume SATA drives that are slower but less expensive, and generally less reliable. Many arrays can support a variety of drive types, allowing you to implement multiple storage tiers within the same disk array -- allowing data to be migrated between tiers and usually resulting in a significant cost savings for the storage organization.
Consider how a disk array will impact backup processes. Most IT organizations are used to tape-based backups using conventional backup software, so the introduction of a general-purpose disk array often causes unexpected disruptions to the backup process, forcing backup administrators to use new tools and adjust procedures to achieve adequate data protection. For example, snapshot and replication tools are significantly different than conventional backup software. Understand that internal changes will be needed to support new disk array products. Note that virtual tape libraries (VTL) are disk arrays that are specifically designed to emulate tape libraries, easing the impact on backup processes. VTL systems are covered specifically in the next chapter.
Consider the impact of rebuild times on the backup process. Failed disks are replaced and their contents are then automatically rebuilt from the corresponding mirror disk or RAID group. However, the rebuild process takes time, which impacts production use of the array along with any backups scheduled during the rebuild time. You may not be able to access the affected RAID group (or access it at significantly lower performance levels) until the failed disk bas been rebuilt. Understand how to respond to disk faults and know how that impacts the backup processes. Advanced features like "pre-emptive rebuilds" can monitor a disk for signs of impending failure and start the rebuild process before a disk actually fails -- minimizing the rebuild time.
Look for data compression technologies to extend disk capacity. As data volumes increase at 60% or more per year, the cost of new disk capacity can be a significant expense. Compression techniques are emerging to fit more data on less disk space, lowering the cost of additional storage. Conventional compression basically works by removing redundant data from each file -- making the files smaller and using less space. Data deduplication, also called commonality factoring or intelligent compression, provides single-instance storage, so only one copy of a particular block or file is actually stored on the array. For example, rather than saving 10 copies of a 100 MB PowerPoint presentation in a backup process, only one copy of the presentation is committed to disk. Note that content-addressed storage (CAS) platforms typically employ data deduplication, but the technology is extending to other disk arrays. CAS products are covered in a later chapter.
Consider the disaster recovery implications of disk backups. Arrays typically support hot swapping between disks, but the disk drives themselves are generally not easily transportable. That is, you wouldn't pop out a 600 GB SATA drive from an array in New York City and send it to a storage facility in Des Moines, Iowa. Consequently, disks are on-site assets and subject to fire, flood and other disasters. Disk arrays are often protected with supplemental tape backups -- dubbed disk-to-disk-to-tape or D2D2T -- or replicated remotely to a secondary array across a wide area network (WAN).
This was first published in November 2007