Are you ready for MAID technology? All disks spinning all the time? Not necessarily.

Given the continuing economic slump for the technology sector, identifying new and potentially disruptive developments in the storage industry has become more difficult than in the past. However, the recent wave of activity and positioning regarding SATA (Serial-Advanced Technology Architecture) disk drives is one of the more interesting concepts developing in the storage industry this year. Some estimates indicate that ATA class disks are expected to be used in as much as 40 percent of all disk subsystems by 2006. The prior generation of Parallel ATA disks use a 100MB/sec shared sub-channel where as SATA disks offer a bus speed of 150MB/sec on a dedicated path using a point-to-point topology The follow-on SATA II drive expected to appear sometime in 2004 will offer a 300MB/sec bus speed. SATA disks are just now being positioned as a new level in the storage hierarchy between online disk storage and Nearline or automated tape library storage.

Between Disk and Tape

Nearline, originally created by StorageTek in 1988, has traditionally defined the level of storage between disk and Farline or shelved storage by using robotics to retrieve storage media automatically. In addition to being the backup medium of choice, Nearline storage today is the primary repository for much of the world's reference data and digital content, and contains between 80 percent and 90 percent of the world's digitally stored data. Far-line, or manually retrieved storage from the shelf, still represents the vast majority of the world's analog or non-machine readable data, as the fast analog to digital conversion is slowing but not halting its rate of growth. Though automation resolved many troublesome issues regarding removable media storage, the time to recover data from tape and the fact that tape only enables sequential access remain as obstacles for some applications. The emergence of the low-cost, relatively high-performance SATA disks are now beginning to address these remaining issues.

This new emerging ATA-based level is being referred to as Near-Online storage or Inline storage and delivers very low price-per-GB storage using SATA drives in high capacity 100TB+ subsystems where all data remains online at all times. Fixed content or reference information including documents, digital images, video, email, medical records, some backup/restore requirements and a wide variety of digital assets are becoming the primary applications for this emerging layer of storage. Most SATA activity and thinking is currently targeted for non-mainframe computing systems.

Introducing MAID

Evolving in parallel within the SATA movement is the new concept of MAID (Massive Arrays of Inactive Disks) storage. MAID is similar to the RAID concept except that in a MAID storage array, all disks (currently SATA disks) are not spinning all the time. In a MAID subsystem, many of the disks remain dormant (powered off) until requested. Power up time for SATA disks takes about 10 seconds. MAID is aimed at enabling the current SATA activity to handle an additional level of storage requirements currently being addressed by automated tape libraries or not being cost-effectively addressed at all. If this sounds somewhat analogous to an automated tape library, with the exception that disks are substituted for tape cartridges, you are headed in the right direction. The MAID concept uses the long-standing storage industry observation that approximately "80% of the activity goes to 20% of the physical storage." By reducing the number of disks that are concurrently active, the disk controller costs can be significantl y lowered by significantly reducing the number of I/O path connections, cache requirements, and overall controller complexity. MAID provides traditional levels of RAID data protection capability for higher availability. The use of MAID storage will be application driven and isn't suitable for all applications, but addresses backup/recovery, lower activity reference data and fixed-content data cost effectively while meeting performance requirements. One company, Asaca, has recently announced a product m the MAID category.

Reference Data Characteristics

Reference data is now growing significantly faster than traditional storage applications and reference data doesn't change much after it is created. One of several reasons for this growth is the new and growing list of government and legal regulations that mandate long-term data retention policies. As an industry; we are deleting data at a slower rate than ever before mainly due to these legal requirements. Maybe the next generation of computer keyboards will even come without a Delete Key. The longer lifetime requirement for digital data retention has triggered much of the current Data Life Cycle Management activity in the storage industry. After reference data ages beyond its active or period of re-reference (suggest that 90-120 days is a good starting point), the data still needs to be retained and automated tape library storage remains the optimal choice and lowest overall cost solution for long-term archival storage. This partial list of regulations is becoming increasingly important to the storage admin istrator's strategy and includes:

* The Sarbanes-Oxley Act requires that defines rules for falsification of records and e-mail along with retention and deletion guidelines.

* HIPAA: Health Insurance Portability and Accountability Act for medical images and records.

* Brokerage Business: SEC Rule 17a-3 and 17a-4.

* Telecommunications: Title 47, Part 42.

* Banking: OCC and FDIC regulations.

* Defense: DOD 5015.2 regulation.

* Numerous others under review.

Rich Media Characteristics

The long-term growth opportunities for reference, content and rich media management solutions are increasing. Rich media consists of voice, text, graphic images, audio, HDTV, 3-D graphics, and movies. Like reference data, rich media doesn't change much after it is created. New and emerging digital applications will continue to fuel many years of explosive growth for storage as terabyte-plus data warehouses, VCR to HDTV quality movies, electronic voice and video-mail, digital security systems, and digital photography all continue to drive major increases in data rates and storage demand. A single movie production can require over 800 terabytes to complete before it is released. The worldwide commercial video storage market including television, movies, commercial production, and video distribution is projected to reach over 700 petabytes (700x10 to the 15th power) by 2006. In the next few years, backup, replication or mirroring techniques will double or triple the original storage demand pushing the rich conte nt market into the exabytes. Rich media and digital content requires much larger multi-media objects (voice, text, video, online games, and high-density graphics) and has pushed storage management requirements well beyond the capabilities of available management tools and only adds to the mounting storage management challenge.

In addition to digital data, a variety of non-digital data applications remain including analog audio, analog video, many types of historical scientific data, books, maps, old music, photographs, and medical images. Paper is estimated to generate over 200 terabytes of digital storage per year, X-rays between 15 and 20 terabytes per year, MRI's remain an unknown, and music downloads onto CDs anywhere upwards of 50 terabytes per year. College campuses in the United States recently reported that on average 75 percent of their campus bandwidth is now consumed from students downloading music.

What Data Goes Where?

The non-mainframe storage industry is leaving a nearly ten-year period when the safe and easy storage strategy was to put all data on disk and back it up to tape. The traditional storage hierarchy is now re-inventing itself based on the emergence of new, lower-cost storage solutions, a need to provide almost instant data recovery, and an entirely new set of parameters governing archival data retention. The role of an HSM type of software product in the Unix, Linux and Win2K markets should begin to accelerate. The economics of this new model is beginning to encourage a multi-level data storage and migration strategy The first level of disk storage remains for online, high-performance, mission critical data. Rich media applications are not typically mission-critical and are being targeted for the Near-Online storage, Inline, or lower cost disk storage.

Conclusion

The emergence of disk-based backup and archival storage solutions is underway. This opens a relatively unaddressed market for disk and tape companies. Led by lower priced SATA disk drives, the role of disk storage is expanding beyond traditional applications. Government regulations requiring much longer-term data retention accelerate the demand for larger capacity tape technologies. The introduction of MAID is positioning itself to successfully bridge the gap between disk and tape storage. Old MAID is a game, new MAID is becoming a reality.

www.asaca.com

www.storagetek.com

RELATED ARTICLE:

Primary disk - Online, OLTP, database, mission critical data

Near-online disk - Fixed content, rich media, fast backup/recovery, and replication

MAID disk - Backup/recovery, reference data, mid-term archives

Automated tape - Backup/recovery, long-term and permanent archives