Sun's ambitious pursuit to steal third-party storage business from EMC
Will Sun hit or miss in its drive to sell storage to non-Solaris platforms? What advantages does Sun offer over the competition?
Sun Microsystems, tired of watching vendors like EMC Corp. and Data General Corp. selling RAID storage products into its SPARC installed base, struck back late last month by launching its first fully-redundant RAID (Redundant Arrays of Independent Disks) offering, the RSM Array 2000. But this wasn't just another storage product introduction -- Sun sees the RSM 2000 as its initial step into the high-end, third-party attached storage market -- a rapidly-growing market Sun pegs at $20 billion this year.
For Solaris customers, the RSM 2000 is an appealing proposition. With controllers optimized for Solaris, the product can scale to 11 terabytes of storage and sports a list price that computes to less than 50 cents per megabyte. In non-Solaris environments, however, Sun's value proposition is less obvious. Right now, Sun has some interesting ideas about how it will knock off market-leader EMC, but there will be some long nights in Menlo Park before any of them become a reality.
What are Sun's chances?
Sun chose its recent high-profile Enterprise 10000 (code-named Starfire) symmetric multiprocessing (SMP) server launch (see our news story "Gunning for the data center: Sun's Starfire targets mainframe co-existence") to introduce the "family" of enterprise storage products that will comprise its first assault on the third-party storage market. The family is composed of the RSM 2000, the Sun Enterprise Tape Library 4/1000 digital linear tape (DLT) subsystem, and the Enterprise Storage Manager Console, which is a software product designed to link up storage, security, and network management applications using Sun's Java Management APIs (application programming interfaces).
Sun is drawing on third parties for all members of the family: the RSM 2000 is from Colorado's Symbios Logic, and the tape hardware comes from ATL Products Inc. Sun will also be working with a variety of third parties, including OpenVision and Legato, to develop software that interoperates with the Console.
Sun will ship the new storage products for the Solaris environment first, with Windows NT and Hewlett-Packard support pledged for mid-1997. And Sun is ambiguous about mainframe support, saying only that it will be available "later." A mainframe offering would not use the SCSI or fiber interfaces that Sun is presently using. Sun says it would also need new software to emulate the mainframe disk format.
The market leader in third-party storage for both Unix and mainframes, Massachusetts's EMC, doubts that Sun will compete outside of Solaris. Seth Traub, senior market analyst with EMC says bluntly, "They don't have any capability to support a non-Sun environment because they have no support services or engineering services trained to do so."
Tom Lahive, an analyst with the market research firm International Data Corporation (IDC), agrees that while Sun might have a third-party strategy on paper, "they don't have the sales or support organization that will really support that non-captive strategy." And even when Sun gets its sales and service in line for other platforms, Lahive says support for NT and HP always will take a back seat to Solaris.
Right now, even Sun admits that multiplatform service is not there. Ed Turner, Sun's storage marketing director, admits that bringing SunService up to speed is "one of the reasons that we've made the availability of HP-UX and Windows NT out in mid-year." Sun says it will be hiring new engineers and sales and service personnel, as well as training its existing sales and service forces. These things take time.
Still, if a company the size of Sun wants to sell to heterogeneous environments, it will try. And, while EMC may have a reputation for outstanding service, Lahive feels that Sun is fully capable of leveraging its well-branded SunService name to make service a strength rather than a weakness.
Sun's strategy is first to take away from third-party sales in the Solaris environment and then sell into NT and the various Unixes, with mainframe connectivity coming last. Turner compares it to eating an elephant, saying Sun is taking a "one-bite-at-a-time approach." "The reason that we targeted EMC in this announcement is that they have been making a lot of noise in our market over the last year or so," he notes. Sun claims 65 percent market share in the Solaris environment. Analysts say EMC's part of the remainder easily is worth $150 million. And while Sun's service story may not yet be ready, the company has some strengths that it hopes will help it knock off both EMC and other third-party vendors.
Sun's fiber and software advantage and its
race against EMC
For the past two years, Sun has been investing in Fibre Channel Arbitrated Loop (FCAL) disk interface technology. With a data rate of 100 megabytes per second, Sun says FCAL is five times as fast as a traditional SCSI interconnect and can make far more host connections as well. It also is touted as having full-duplex capabilities (the ability to do read and writes at the same time) and being easier to hot-connect than SCSI.
Sun is reluctant to mention specific product plans, but it has been shipping fiber products in its SPARCstorage line for more than two years and claims to have the largest installed base of fiber channel storage in the industry -- 1.5 petabytes installed to date. Turner doubts that Sun will be first to market with direct-interface fiber storage, but he clearly would like to beat EMC, which is preparing a similar product. "We will be early to market certainly," he says, and "it wouldn't surprise me if we were the first volume shipper of fiber channel drives."
The Holy Grail for both companies right now is to ship storage products that use fiber all the way from the host to the array to the drive. Current SPARCstorage fiber products use SCSI for the array-to-drive connection with clock data rates closer to 25 megabytes per second than the 100 megabytes per second Sun showed off with direct-interface fiber at Comdex the past fall.
Using Java to gain ground
Perhaps the most significant way Sun is trying to differentiate itself in the third-party storage market is through its use of the Java management APIs, which are being integrated into the Enterprise Storage Manager. Here, Sun's plan is to link diverse storage management applications like OpenVision's NetBackup and Enterprise HSM within a Java-enabled console so that all applications can run through the same interface.
With Enterprise Storage Manager, says Turner, "if you put our RSM 2000 on an HP machine, you will be able to manage that from another terminal in your network." Because the console is written in Java, management can be done via TCP/IP anywhere there is a Java virtual machine. IDC's Lahive says this integration of storage management utilities is "an enormous competitive advantage" for Sun because, as the job of the storage manager becomes more complicated and the number of files being stored continues to skyrocket, "anything you can offer them is advantageous -- especially at the enterprise marketplace."
The drawback to the Java APIs is that they do not have the universal support that would make them truly useful in heterogeneous environments. To date, only Sun's OEM partners, including Legato and the soon-to-be-merged OpenVision/Veritas companies, have promised to write to them; none currently is shipping software. Sun itself can't say for sure when it will be shipping the console. When it does, managers will have to replace their existing management software with Java API-enabled code -- a costly prospect.
Sun or EMC on price performance -- one company's view
To David Flaxman, chief scientist for electronic commerce at KPMG Peat Marwick LLP, Sun's initial value proposition is simple: Sun looks less expensive. He believes that when it comes down to the RSM 2000 versus EMC's new Symmetrix line, "what you're likely to find is that Sun will have significantly better price performance." He adds, "even if Sun doesn't quite hit the level that EMC occupies -- and maybe they will -- they are undoubtedly going to be offering an alternative between commodity storage and incredibly high-end storage."
Flaxman thinks it is possible that Sun's products could work well in a heterogeneous Unix environment and says he could live with or without mainframe connectivity from Sun. "Usually, Unix and mainframe environments are well separated. I think that getting affordable Unix storage is an important enough goal that the market will get behind it."
The one certainty is that Sun stands to make money from its new family of storage products. Presently, IDC estimates the Solaris storage market at close to $1 billion, and everyone seems to think that Sun will increase its presence there. Whether it will be able to cash in on the $6 billion Unix market, or even the $20 billion worldwide storage market, depends on whether it can commit itself to platforms beyond Solaris.
If you have technical problems with this magazine, contact firstname.lastname@example.org