In-Depth

View from the Crossroads: Can iSCSI Make Storage Management Friendly?

A widely respected storage veteran looks at connectivity issues, including SCSI/iSCSI, plus Microsoft and storage management.

• By Jon William Toigo
• 03/23/2004

In the previous column, covering “SCSI family counseling” solutions (that is, bridges, routers, and gateways designed to bridge one implementation of SCSI to another), we quoted from a phone interview with Bob, who is Vice President and Chief Technologist for Crossroads Systems in Austin, TX. However, the brief citation didn’t begin to do justice to the lengthy conversation and the many important insights he offered in his interview.

First, a brief background. Griswold is called Chief Technologist rather than Chief Technology Officer probably because Crossroads has cut back in recent months on the number of corporate officers on its masthead. For all intents and purposes, Griswold is a CTO: not just the airy visionary variety, but, we suspect, the sleeves-rolled-up, arms deep-in-the-racks type with more than a few cuts and calluses to show for working around the sharp tin enclosures.

Over his career, Bob has worked for most of the brand name storage vendors—and for more than a few smaller firms that were gobbled up by the bigger fish. He is pedigreed and respected in the industry, but more important, he is willing and able to answer the “why” questions that most marketing and engineering folk eschew.

On the subject of Crossroad’s primary business, which is building bridges between different storage channel protocols, he has pithy statements to offer about competitors. He offers that ATTO Technologies has legitimate claim to offering the first parallel SCSI-to-iSCSI bridge, “They demo’ed their first generation iSCSI product it at the Spring 2003 Storage Networking World Conference—a 4u, 6-by-6 board.”

He quickly adds that ATTO faced challenges in terms of pricing for that first product. “The OEMs were looking for that kind of technology to take advantage of the standardization of iSCSI and its potential appeal in the [small and medium business] (SMB) market. Their original Manufacturer Suggested Retail Price (MSRP) was around $4995, which was three or four times what most OEMs were looking for [to deliver solutions within the price sensitivities of the SMB market while still earning profit on the solution]. They are a private company, so it's hard to get good financial data, but the product didn’t seem to have been a resounding success.”

Crossroads, he says, should have an SMB-friendly product available this year that aligns with both customer requirements and OEM and integrator objectives. He adds that it is no easy task to bridge SCSI and iSCSI, especially when the target device is a tape subsystem. He explains, “Many people don’t understand the sensitivity of tape to continuous streams of data. It isn’t simply an issue of stripping off the iSCSI or Fibre Channel packaging layers and delivering the data and instructions to the tape device. There is a lot more to it than that.”

He points to Boulder, CO-based SpectraLogic as an exception to the rule when it comes to iSCSI tape, “They are unique in that they do all of their connectivity engineering in house. It is no surprise that they have their own SCSI-to-iSCSI bridges inside their libraries. They were also among the first to offer NDMP over Gigabit Ethernet. They’ve also done a great job of marketing their own smart library operating system, TAOS.” Fortunately (if you are Crossroads) or unfortunately (if you are a consumer), most other tape vendors haven’t gone as far as SpectraLogic in adding intelligence or connectivity to their products.

Griswold says that many big iron companies—tape vendors and disk array vendors—do not understand the SMB market. “For SMBs, there are two kinds of costs. Hard costs are the costs to purchase and deploy technology, which they understand. The other kind of costs are unknown costs—‘How many classes do I need to send Roy to before he understands how to set up a zone in a Fibre Channel SAN?’ That is a big problem for the Fibre Channel world as they try to sell products to the SMBs. Many enterprise SAN product vendors just don’t get this point.

“With iSCSI, there is an opportunity for the industry to play into the sweet spot—the SMB market. They have money to spend and they will spend it on iSCSI SANs if they can see that it will be simple to deploy and manage. If you could just get everyone to play along so that any HBA, any switch, any controller could plug in, then everyone could make money. A lot of vendors don’t get that point.”

The potential for iSCSI, he argues, is that it could make storage more “management friendly.” He notes that there is considerable concern among many storage resource management software (SRM) companies today regarding Microsoft’s support for iSCSI, which he describes as “a fear of OS encroachment on their functional space.” At the end of the day, Microsoft, Intel, and Dell will drive storage management requirements in the SMB space, he says, “because these are mainly Windows shops.”

He adds, “There is a lot of fear that Microsoft will add storage management utilities directly to its server OS. They are already beginning to do virtualization services. They will probably take whatever comes out of ANSI’s T11.5 Task Group, where the T11.5 is working on [the definition of device level interfaces between management applications and all devices in a storage network], and use it to extend their control over storage resources. I have been watching the T11 folks that are writing the FAIS spec, and with no actual code linking that effort back to the SCSI (T10) standards, it’s success, at least stand-alone, is up for debate. It’s an open door for Microsoft and other OS vendors to simply take the lead and to decide what resources are available to their servers and how they will be allocated.”

Griswold paints a picture of the SMB storage environment in the next few years: “I see SAS/S-SATA becoming the dominant internal array architecture because it leverages well-understood technology at the right price point, and iSCSI becoming the dominant SAN interconnect because it leverages existing LAN switches. And I see Microsoft becoming a dominant force in the management of the platform.”

To his way of thinking, it’s all about the Benjamins. What do you think? E-mail me at [email protected]