SAS Protocol Rounds the Final Bend

At the third and final scheduled Serial Attached SCSI (SAS) Plugfest, held the week of November 1 at the University of New Hampshire, it was clear that products incorporating the SAS protocol are edging closer to market. In fact, Henry Mason, director of industry marketing at LSI Logic and president of the SCSI Trade Association (STA), anticipates that SAS products could begin to appear in the indirect sales channels before the end of year and from major storage OEMs in the first quarter of 2005.

Should Mason's prediction prove accurate, the development would bring to fruition a 3-year effort to provide storage administrators with options that fit squarely into the emerging concepts of Information Lifecycle Management (ILM). Perhaps just as important, the standards-development process could serve as a model for development efforts on future standards.

The STA standards group conceived SAS as the next evolution of the long-established and long-dominant SCSI protocol, and designed it to meet the high performance requirements of enterprise data centers. But perhaps SAS's most distinctive feature is that it's a superset of the Serial ATA (SATA) protocol. For this reason, both high-performance storage devices and lower-cost ATA alternatives can use the same protocol. Over the past several years, ATA disk drives have grown in popularity, particularly for near-line or less-demanding storage applications. According to Marty Czekalski, interface architect at Maxtor and vice president at STA, administrators who use SAS will be able to use the same tools and protocol stacks to manage both high-performance and lower-performance storage technologies.

This capability fits neatly into the emerging discipline of ILM, in which storage administrators distribute data across different tiers of technology depending on its perceived value. With SAS, administrators will be able to leverage hardware investments, mix and match components, and potentially reduce the number of components in the hardware infrastructure. As a result, Czekalski said, support costs should go down.

As significant as the emergence of the protocol itself, however, is the success of the standards-development process. SAS development efforts began 3 years ago, and products incorporating the protocol will appear almost exactly on schedule, if not a little early. The timeline has held up very well, and the overall plan remains on track.

According to Mason, the effort was successful for several reasons. Perhaps most important, the group's objectives were very clear. First, the group wanted to integrate the SAS effort with the roadmap that was already in place for SATA. Consequently, the SAS development group borrowed elements from SATA to ensure compatibility. Second, the group has a strong preservation objective. It was committed to maintaining the SCSI legacy software. Third, the group lined up strong OEM commitment. Vendors wanted to introduce new products, and that commitment provided an incentive to continue to move the specifications forward within a reasonable timeframe. The standards team was determined not to invent specifications for the sake of invention but to borrow whatever could be borrowed from Fibre Channel and SATA. Finally, the desire for a common platform and infrastructure for different classes of drives resonated within the industry. In the past, the different flavors of storage technology have driven support costs up sharply.

But clear objectives and widespread industry support were only part of the formula for success. Many of the champions of the SAS protocol had experienced the challenges of previous standards efforts, and one of the major changes they made was in the use of the plugfests. In every other standards effort, the plugfests were central to figuring out why things didn't work after the development was complete. In this effort, the plugfests were part of the process and integral to the development effort. Problems were identified sooner rather than at the end.

Twenty-one companies participated in the most recent plugfest, in which interoperability among components and other pieces of the puzzle were tested. Although the results are confidential, nothing emerged to lead Mason or Czekalski to believe that the timeline will change. And as products begin to emerge, both men anticipate that adoption will be rather swift. Nonetheless, they caution, people who purchase the technology still must observe due diligence to ensure that the choices work as anticipated and meet their needs.

In the final analysis, the emergence of SAS-based products will mark a major milestone in the storage industry. Not only is the technology applicable to the emerging direction of storage, but it also reflects a successful standards-making process that can be replicated in the future.