Pillar Leaves Post - At Last

After spending nearly four years and $150 million in development, Pillar Data Systems has a product to ship.

Until now, the company has been known primarily as Larry Ellisons storage toy (see Larry's Stealth Storage Startup). The Oracle Corp. (Nasdaq: ORCL) CEO is Pillar’s lone investor through his Tako Ventures firm, and he put up the $150 million that let the startup grow into a 350-person organization headquartered in San Jose, Calif., before finishing its product.

Now -- drumroll -- enter the Pillar Axiom 500, an integrated NAS and SAN system for the midrange. Is it worth the wait since 2001?

“It has been a long time,” Pillar CEO Mike Workman admits. “We didn’t spend $150 million playing ping pong. Larry wants a storage company, and he’s interested in a home run, not a bunt. This product had to be feature rich.”

Still, at this stage, Pillar seems to have left more features on the roadmap than it put in its product. Pillar is shipping NAS, but the SAN part is mostly still in beta testing with limited customer availability until later this month. It supports SATA drives, but Fibre Channel drives are six months out. So is iSCSI support -- and some advanced software features such as replication.One thing: After nearly four years on the QT, Pillar has developed a lingo of its own. Its hardware components consist of slammers, bricks, and pilots. No, a slammer isn’t where bad data is sent for punishment. In Pillar-speak, it’s a controller that can be set up to connect to disk via Fibre Channel for SAN or Ethernet with NAS. An Axiom system supports up to four slammers, connected via Emulex Corp. (NYSE: ELX) embedded switches.

Bricks are RAID enclosures, and pilots are controllers that host management software.Workman says he’s especially proud of Pillar’s software, which consists of 2.5 million lines of codes.

Besides management wizards to simplify setup, the software manages system resources to provide multiple quality-of-service (QOS) levels from a single storage pool. The software provides configuration of data placement on disk, spindle striping levels, mirroring, priority queues, network bandwidth, and caching algorithms that can match up with specific applications. For instance, database applications that require high QOS could be stored in the outer edge of the disk, where data rates are higher, and lower priority data could be stored in the inside tracks with lower data rates.

It remains to be seen how that will work, but for now Pillar appears to be following Network Appliance Inc.’s (Nasdaq: NTAP) strategy of offering NAS and SAN in the same system. Workman says there’s a difference: NetApp started out as a NAS filer and added SAN capability, while Pillar’s system was engineered to support both from the start.

Be that as it may, at least one customer likes what he sees so far. Chris Hill, associate director of information systems for global law firm Thacher Proffitt & Wood LLP , bought an Axiom after running a test system with SAN and NAS. Hill says he uses it to run Microsoft Exchange at the highest-priority QOS; SQL Server at medium QOS; and general storage at low QOS. Hill also expects to eventually place SATA and Fibre Channel in the same system, which allows him to put 12 Tbytes of storage into a 15U box.He’s still waiting for other promised features, such as support for EMC’s VMware ESX server virtualization software and advanced replication (Pillar is temporarily using NSI Software Inc. for replication). He doesn’t mind that Pillar doesn’t support other vendors’ hardware but acknowledges that could concern others.

Hill also likes the price, $100,000 for 12 Tbytes: “This system seems to target a real sweet spot that’s missed by other storage vendors. It’s unusual for us to buy from a startup. But it’s such a departure from the existing storage paradigm that it’s worth it to take a leap.”

Analyst Jack Scott of the Evaluator Group says Pillar’s financial setup makes it different than most startups. “They’ve got a little more substance than other startups,” Scott says. “They have much deeper pockets than anybody else I know.”

Workman says he doesn’t want to test the depth of Ellison’s pockets. Although he won’t say how many customers he has -- “not enough, we’ve only been selling 30 days” -- he says he expects to be profitable sooner than people expect. Workman says the goal is to become a large independent company, fast.

Scott says Pillar has a better chance than most startups, but not even Ellison’s money can alleviate some of the challenges it faces. “It’s real tough for startups to become significant,” he says, “A lot of the big folks don’t want to buy disk from Pillar Data. They want to buy from EMC or HP or somebody like that.”— Dave Raffo, Senior Editor, Byte and Switch