In-Depth
Going After the SMB Market
Why does an SMB need—or want—a SAN at all?
I have to confess that I cringe every time I hear the words “SANs for the rest of us” issue from a vendor’s mouth. The expression is a code phrase for Fibre Channel fabrics for the small-to-medium-sized business (SMB)—a market segment notorious for its reluctance to embrace FC technology. Since the SMB market is also at present the only segment of the consumer market demonstrating meaningful growth in technology purchasing, it is no wonder that FC product vendors would do their best to curry SMB favor.
Last year, I had the opportunity to participate as a keynote speaker for a multi-city Network Storage University road show in which QLogic and Microsoft were among the many vendor sponsors. City after city, I listened to the speakers, mostly from QLogic, talking about how they were working together to build a plug-and-play Fibre Channel SAN to support SMBs. An FC switch and HBA vendor, QLogic was touting the relationship as the harbinger of “SANs for the rest of us.” Shortly, SMBs, which are predominantly Microsoft shops, would be able to roll out an FC SAN quickly and easily to support their burgeoning data.
In conversations with Microsoft, I got a very different impression. Redmond’s interest in Fibre Channel was mainly part of an effort to reinforce the impression of their server platforms as “enterprise class.” Microsoft already owns the SMB market. Their FC play was designed to add appeal for their products in the world of Big Iron—the upper tier of the market.
So, I concluded that it was simply a matter of each vendor putting its own spin on the facts to support its particular marketing objectives. Nothing to write home about.
However, Emulex announced that it, too, is making a play in the SMB space. According to Executive VP Mike Smith, at Emulex headquarters in Costa Mesa, CA, the company is launching a new product —the LP101 adapter—targeted directly at the SMB consumer. He gave much the same argument as did QLogic (but without the Microsoft connection).
“OEMs are pushing us to provide products in this space, and we have focus groups that have indicated a readiness of SMB shops to centralize their storage,” Smith explained. “We think that shops with 8 to 10 servers are experiencing problems with allocating disk and doing backups that are driving them to adopt SANs.”
Emulex, of course, wants to help the poor little guys with a Fibre Channel HBA that deploys simply and easily using a configuration wizard. “Our high end FC HBAs have tremendous flexibility, with many setup parameters … that enable the card to be tuned and optimized to meet the needs of an enterprise environment. But this complexity, which is appropriate to the sophisticated IT manager or storage manager in the data center, may be a burden to the smaller user. So we are hiding and eliminating the complexity in the setup [of the LP101] and enabling the configuration of the HBA via a couple of clicks.”
Smith says Emulex has no preconceived notions about the appeal of FC SANs to SMBs. However, citing data from Gartner Group (which could be, arguably, dubious), he notes that FC HBAs enjoy a presence in the top 5 to 10 percent of servers (e.g., in large enterprise shops), while 60 to 70 percent of servers continue to use HBAs in connection with direct attached configurations. He acknowledges that DAS remains the dominant storage topology, especially in the SMB market. However, he says that with even a “conservative penetration” by FC HBAs and FC SANs in the SMB space, the rewards to the successful vendor could be as high as $250 million in additional revenue. No small potatoes, those SMBs.
Alternatives to FC SANs for the Rest of Us
Emulex, Qlogic, and others confront a crowded group of contenders as they vie for market share among the SMBs. Alternatives to FC SANs include iSCSI (SCSI over IP networks) and Serial-Attached SCSI (SAS), the next evolution of the traditional SCSI protocol.
Backers of iSCSI say the protocol is easier to understand than FC and uses familiar IP networking gear. While iSCSI is less efficient than FC (using only 70-80 percent of available interconnect bandwidth), this limitation is less painful in an SMB shop than it would be in a big, transaction-heavy enterprise setting, advocates claim. (The point is pretty much rendered moot when 10GigE IP switching becomes prevalent in these settings as well.)
Smith offers that while iSCSI SANs could compete for SMB attention, “Those products are clearly not ready for prime time yet.” Tell that to SANRAD, Adaptec, LSI Logic, Alacritech, EqualLogic (among others) with iSCSI products in the market today.
Smith wisely adds that he does not imagine that “Fibre Channel is going to dominate the SMB market.” Given the ascendancy of Serial-Attached SCSI (SAS), it is probably safe to assume he's right.
Serial-attached SCSI has the advantage of being a minor upgrade to the parallel SCSI bus-based configurations most shops use today. However, SAS improves on traditional SCSI by its support for greater distances between servers and storage platforms (six meters per connection) and support for many more devices than parallel SCSI could handle. With “expanders,” even small companies could build pretty impressive infrastructures that combine up to 16,000 SAS and/or serial-ATA devices with interconnect speeds of up to 600 MBps.
Bottom Line
The number of alternatives for building storage area networks (iSCSI) or quasi-networks (Fibre Channel and SAS) is increasing. And, because of their purchasing strength, vendors have painted targets on the backs of SMB IT professionals. The unanswered question in all of this hullabaloo over SANs for the rest of us is: Why? Why does an SMB need or want a SAN at all?
My advice: Don’t let anyone sell you a SAN until they can make a real business case for the technology that considers your applications and business requirements. Such advice is probably unnecessary. Given their cost sensitivity and common-sense mentality, SMBs tend to be among the best and brightest of storage consumers.
About the Author
Jon William Toigo is chairman of The Data Management Institute, the CEO of data management consulting and research firm Toigo Partners International, as well as a contributing editor to Enterprise Systems and its Storage Strategies columnist. Mr. Toigo is the author of 14 books, including Disaster Recovery Planning, 3rd Edition, and The Holy Grail of Network Storage Management, both from Prentice Hall.