Mike Workman, CEO, Pillar Data Systems

For years, people had only one question for Pillar Data CEO Mike Workman: What was his startup doing with Oracle founder Larry Ellison's money?

About a year ago, Pillar set out to answer the question. The startup that Ellison funded and pumped $150 million into before it delivered a product began shipping its Axiom 500 system last June. (See Pillar Leaves Post At Last.) Workman, who spent most of his 30-plus years in the storage business with IBM and took over as Pillar CEO in 2001, is happy these days to have a product to talk about instead of getting grilled about wasting the boss's money. Although he likes to point out that his boss wasn't among those doing the grilling.

Figure 1: Mike Workman, CEO, Pillar Data Systems

"For a long time everybody commented that it took us three-and-a-half years to come out with a product," Workman says. "'What did you spend $200 million doing?'

"We sat down with Larry and said, 'It's hard, lots of code, lots of enterprise code, serious stuff.' We spent six months testing once the code was done. Larry looked at what we were doing and said, 'It's hard to believe you have done what you've done so far.' He has an appreciation for enterprise stuff. Everybody else said, 'What the hell is taking so long?'”Pillar's mission is ambitious. (See Larry's Stealth Storage Startup.) Its systems are designed to be used for SAN and NAS, three tiers of storage, and everything from transactional data to backup and archiving -- all in one box. And while many startups tackle a new technology area that pits them against other private companies, Pillar is taking on established storage leaders such as EMC, IBM, Hewlett-Packard, Hitachi Data Systems, and Network Appliance.

"Three million lines of code later…it's tough to develop one platform that goes against a lot of products all in the same box, with the same storage platform, same management software," Workman says. "We range in price from $75,000 to $1 million. We do go directly against just about all the products from EMC, IBM, HP, Hitachi, and NetApp. Well, we don't go against the [low-end HP] MSA 1000 and [EMC] Clariion 300…yet. We'll have a whole new box coming out for the midmarket."

We recently sat down with Workman to find out exactly what he has been doing with Larry's money, and how he plans to spend more if it before Pillar becomes profitable.

— Dave Raffo, Senior Editor, Byte and SwitchContents:

NEXT: Great Expectations

Byte and Switch: How many customers do you have?

Workman: 87.

Byte and Switch: Was that your target for this time when you started shipping?Workman: I think we had higher expectations, which were not well founded. [Laughs.] The nice thing about being a startup is you're not burdened by the shackles of past success, because you don’t' have any.

Byte and Switch: After spending nearly four years and $150 million in funding, didn't you have more shackles than most startups?

Workman: For a long time everybody commented that it took us three-and-a-half years to come out with a product. "What did you spend $200 million doing?" But we covered the broad spectrum from VTL [virtual tape libraries] to running transactional oriented applications.

Three million lines of code later... It's tough to develop one platform that goes against a lot of products all in the same box, with the same storage platform, same management software. We give people three tiers of storage in one system. We give them file and block storage. We range in price from $75,000 to $1 million.

We do go directly against just about all the products from EMC, IBM, HP, and Hitachi. Well, we don't go against the [low-end HP] MSA 1000 and [EMC] Clariion 300...yet. We'll have a whole new box coming out for the midmarket.Byte and Switch: Did you consider coming out with a less ambitious product sooner and bringing out a higher-end system later?

Workman: Maybe we could have. We considered that. We decided we would stick with our principles. Once you get something in concrete, it's hard to tear it apart. If you go down a tributary off the main river, you might never get back to the main river.

We sat down with Larry and said, "It's hard, lots of code, lots of enterprise code, serious stuff." We spent six months testing once the code was done. Larry looked at what we were doing and said, "It's hard to believe you have done what you've done so far." He has an appreciation for enterprise stuff. Everybody else said, "What the hell is taking so long?"

We tried to stay away from being a niche company as far as what we want to do with the company. Niche companies look to sell to a big fish. The VCs like that, Larry doesn't.

Byte and Switch: When will you break even?Workman: Another year. The hard part is continuing to scale revenues and customers without losing customer satisfaction.

NEXT: Bricks and Slammers

Byte and Switch: Maybe you needed so much time because you came up with new names for the storage devices. You have slammers [controllers], bricks [enclosures], and pilots [management software]. How are they different from other vendors' systems?

Workman: You can configure slammers either as a SAN slammer, a NAS slammer, or both. You can have up to four slammers per system. The difference between a SAN slammer and a NAS slammer is the connection. Our NAS box comes with Ethernet connections, SAN comes with Fibre Channel. SAN customers often configure the box with more cache.

As you scale slammers, you scale the fabric as well. Each brick has two hardware RAID controllers in it, so you add SATA bricks to add RAID controllers. Every time you add 13 disk drives, you add RAID controllers.Byte and Switch: Are customers using you mostly for SAN or NAS?

Workman: About 70 percent are SANs and 30 percent are NAS. About 15 percent of those are both, so it could be 55-45 or 85-15.

Byte and Switch: You're SATA only, right?

Right now we are. This summer we will add Fibre Channel. Also iSCSI.

NEXT: RoadmapByte and Switch: Are you seeing interest in iSCSI yet?

Workman: The whole purpose of the damn thing is to simplify, consolidate, and save. I look at these systems as an adjustable wrench. The thumbscrew is iSCSI. When 10 GigE comes out, that's when I think iSCSI will take off. There will be a serious challenge between FC SANs and Ethernet SANs then.

Byte and Switch: Do you plan heterogeneous support for your systems?

Workman: Probably not. When we started development, people were working on things like StorageTank [IBM] and WideSky [EMC] for multivendor support. (See IBM Virtually in the Game, IBM Gasses Up Storage Tank, EMC Widens WideSky Program, and Users Frosty on EMC WideSky.) Heterogeneous platform management was extremely difficult to develop.

Larry and I talked about it, and he said, "If it takes five years to develop heterogeneous storage management, everything that existed at the point you started is off the books and gone." They [StorageTank and WideSky] cost [IBM and EMC, respectively] more than $200 million [to develop]. It would have cost that much squared if we tried to do a virtualization platform.Byte and Switch: How do customers manage your systems?

Workman: Every box has its own storage resource management built in.

Byte and Switch: It only manages Pillar storage?

Workman: Yes.

Byte and Switch: What about replication?Workman: Pillar to Pillar replication, we do ourselves. Heterogeneous replication will always be OEMed. We use Kashya for block replication, Signiant for file replication. Snapshot software comes form Pillar with all the products.

Byte and Switch: You've talked about a midrange system this summer. Will you go higher, too?

Workman: Eventually we'll go to 64 bricks. It will be on par with the [NetApp] Excelsior product. (See EMC, NetApp Ready New Wares and NetApp Nudges Closer to New OS.)

— Dave Raffo, Senior Editor, Byte and Switch

Organizations mentioned in this article: