Serial ATA, not Serial-Attached SCSI, provides "good-enough" storage and will be the next disruptive technology. By Richard R. Lee
Most IT managers are familiar with author Clay Christensen and his concept of "disruptive technologies." He has shown in numerous case studies how new, "good-enough" technologies can disrupt mature, high-margin industries. The parallels with the storage industry are more than coincidental. A perfect example of a disruptive technology is the emergence of the Serial ATA (SATA) disk drive interface.
For more than two decades, IT users have had to rely on storage solutions that cost orders of magnitude more than the class of storage devices used for desktop, mobile, and consumer applications. Much of this cost disparity has been driven by vendor arguments mandating the use of so-called "enterprise disk drives" (mostly parallel SCSI and Fibre Channel today) that have superior performance, reliability metrics, etc. However, many end users recently have voiced their rejection of this by simply not purchasing additional storage or by better utilizing over-provisioned and under-utilized capacity that they already have. Many users are now open to "good-enough storage" technologies/products.
IT management has determined that storage must have a new value paradigm for them to continue to make significant investments. Given current economic conditions, the storage vendor community finds itself between a rock (continuing to support the status quo) and a hard place (embracing a new value paradigm).
As with any paradigm shift, there must be wide-scale acceptance of the new value proposition for success to occur. For a long time, many IT managers have believed that in order for storage to be "good" it had to be expensive. Debunking this theory is the concept of "good-enough storage," which is characterized by commodity-like prices and products that meet the majority (if not all) of the technical and reliability requirements of enterprises.
"Good-enough storage" provides the best of both worlds, delivering enterprise-level features at commodity prices. One example is the increasing adoption of ATA-based disk systems for data protection, which is challenging tape's dominance in backup and archiving applications. In this case, "good-enough storage" enables companies to move from simple disaster recovery to true business continuity. This is facilitated by local and remote mirroring and replication services with near-instantaneous recovery from errors and outages, with cost structures that are competitive with tape automation solutions.
Another example is in meeting the growing regulatory requirements for archiving and searching the voluminous amount of data associated with electronic business records, medical imagery, etc. Most companies have been addressing these requirements with tape automation because disk-based approaches have been considered to be beyond their budgets. In these applications, "good-enough storage" may prove to be the best approach, providing low cost, high capacity, and acceptable reliability and performance metrics.
Disk drive vendors present two faces to the marketplace—one to their desktop customers and another to their enterprise storage customers. Desktop customers represent high-volume, low-profit opportunities, while enterprise customers represent low-volume, high-profit opportunities. Protecting their enterprise business is a high priority for all disk drive manufacturers, as well as for vendors of controllers, monolithic arrays, etc., all of which have engaged in some questionable marketing tactics recently.
Two of the most telling of these tactics have been vendors telling customers that Serial ATA drives are "not good enough" for enterprise environments and extolling the virtues of Serial-Attached SCSI over Serial ATA in an attempt to protect their current parallel SCSI-based businesses. These efforts are designed to stifle end-user interest in Serial ATA and to keep current and future customers focused on so-called enterprise-class disk drives as the only viable solution.
These efforts to slow the inevitable adoption of Serial ATA systems will not succeed. One downfall of this strategy is a failure to realize that "good-enough storage" solutions are long past due and end users will not wait any longer. We have seen major disruptions by "good-enough technologies" in the server, networking, telecommunications, and application software markets, and the same disruptions will occur in the storage market.
Richard R. Lee is founder and CEO of The Storage Consulting Group (www.storageconsulting group.com) and an editor-at-large for InfoStor.
Disk drive interfaces at-a-glance
For the past 10 years, the dominant disk interfaces have been IDE/ATA and parallel SCSI. IDE drives have been used exclusively in desktop and portable systems, while SCSI (and Fibre Channel) has dominated the enterprise (server) market. IDE/ATA drives represent the vast majority of all the drives manufactured worldwide.
Late last year, the Institute of Electrical and Electronics Engineers (IEEE) ratified a standard for a replacement interface for IDE/ATA, called Serial ATA (or SATA). Serial ATA was designed to overcome numerous shortcomings of IDE/ATA drives (i.e., rotational speed, interconnect distances, I/O rates, total number of addressable devices, and lack of native command queuing). The IDE/ATA market is expected to rapidly transition to Serial ATA over the next year.
Early this year, a consortium of leading SCSI vendors announced a new draft standard—called Serial-Attached SCSI, or SAS—designed to replace parallel SCSI, which has reached its technological limits. SAS incorporates all of the technology found in parallel SCSI drives, but introduces some new capabilities (serial connectivity, a larger number of addressable devices, longer interconnect distance, etc.). SAS also uses the same type of 7-pin connector as Serial ATA drives. SAS drives will be available in 2004 from several drive manufacturers and will be initially targeted at the enterprise market.
While similar in basic attributes, Serial ATA and SAS have one glaring difference: cost. Serial ATA drives currently range from $0.01 to $0.02 per megabyte, while SAS drives are projected to range from $0.05 to $0.08 per MB.