Rich Napolitano, VP of Sun's Data Services Platform Group

"Frankly, I voted with my feet. I didn't have to do the deal."

February 25, 2003

16 Min Read
Network Computing logo

When a huge company acquires a startup, the smaller player often gets sucked into a black hole never to be heard from again.

Not Pirus Networks. It's basically been full speed ahead for the storage virtualization switch crew, which was acquired last September by Sun Microsystems Inc. (Nasdaq: SUNW) in a deal worth around $165 million. Sun left the startup – still based in the same Acton, Mass., offices – largely intact, although it didn't retain Pirus's salespeople (see Sun Beams on Pirus and Sun Completes Pirus Acquisition).

Since becoming a Sun employee, former Pirus CEO Rich Napolitano has been out selling the switch as aggressively as ever. "Really, the job hasn't changed much, to tell you the truth," he says.

This isn't completely true. The Pirus technology, from a marketing perspective, has already been subsumed into Sun's N1 strategy, which is supposed to unify the management of data center resources (see Sun to Pilot N1 in Q1). Meanwhile, the next generation of the Pirus technology [ed. note: not to be confused with Toyota's Prius] will be delivered as an integrated piece of Sun's StorEdge 6900 array. That's definitely a shift from Pirus's previous goal, which was to eliminate a customer's reliance on a single vendor's storage arrays.

Napolitano, however, argues that Pirus's original mission is consistent with Sun's culture of offering open, standards-based systems. It's only that the channels for how that technology is delivered have changed. "What Pirus was about was moving [intelligence] to the edge, which was intended to give people degrees of freedom to pick their back-end storage subsystem," he says. "That's still our proposition." Or to put it another way: Maybe the Kool-Aid at Sun and Pirus is pretty much the same flavor?But Sun has sidelined the iSCSI and file system technologies that Pirus spent significant engineering resources developing. Napolitano, who's now VP of the newly created Data Services Platform Group in Sun's storage division, says his group's top priority now is to find the "quickest path to revenue" and that the market for IP SANs simply isn't ready yet.

"I think we overestimated, ourselves, how fast iSCSI would be adopted," he says, when asked if there was anything he would have done differently at Pirus. "Fortunately, our proposition was broad enough that we could morph – we could deemphasize iSCSI, focus on Fibre Channel, or NAS, wherever the market opportunity is."

Napolitano started his career in high tech at Digital Equipment Corp. After that, he became director of software at BluePoint Technologies, which was acquired by Adobe Systems Inc. in 1989. He then founded Data Kinesis, a hardware RAID and file system controller developer, which he sold to Adaptec Inc. (Nasdaq: ADPT) in 1996.

But he doesn't see another startup in his future, saying he's committed to building Sun's storage business. Besides, Napolitano doesn't want to tempt fate: "You start doing the math of the probabilities [of successfully launching a startup], and it starts getting very, very small of hitting another one."

Read Napolitano's full interview with US Editor Todd Spangler by clicking on the links below:

Byte and Switch: As VP of the Data Services Platform group in Sun's storage division, what do you do?

Napolitano: Really, the job hasn't changed much, to tell you the truth. The charter is to deliver data services – the data services platform, which is the rebranding of the Pirus products under the Sun umbrella. The focus of what we deliver, and when, is being ordered based on the channels at Sun, and how we see the quickest path to revenue, which is really around block virtualization first. So the first-to-market products are around block technology.

Byte and Switch: When is that going to be available?

Napolitano: So it's in customers' hands now. The initial feedback is really, really positive. It's easy to install, performs well, scaleability is good. We're trying to get early reference accounts to really build our base. The technology will come out in two forms of Sun products: the standalone product, which will be delivered through professional services, which is what we're doing now; and then an integrated product, where we embed the Pirus technology into large-scale RAID subsystems. And that's later this year.

Byte and Switch: So that will be embedded directly into the Sun arrays?Napolitano: That's correct.

Byte and Switch: And what's the advantage of doing that?

Napolitano: Basically you can bring higher-end data services to the array. So you bring virtualization capabilities, snapshot, replication services, dynamic multipathing, etc., to a large-scale array... This is in Sun's product line, the [StorEdge] 6900 series.

Byte and Switch: Why not implement those services in software?

Napolitano: For a couple of reasons. You really want to have a lot of scaleable computation, and an array isn't really designed to do that. So if you look at our architecture, what you see is all of our ports have a lot of CPU associated with them. So we can really scale our performance with the connectivity.Byte and Switch: When you say it's able to scale, how do you define that?

Napolitano: We have some number of Fibre Channel ports, up to 32 in the first-gen product, and behind that we have CPUs. The reason why we can scale is that as you add ports, you add CPUs. That allows the storage-oriented applications – snapshot, virtualization, replication, file system technology – to scale as you have more and more connectivity. It's based on the architecture of CPUs on each port, versus ASICs on each port.

NEXT: 'Selling Virtualization Isn't the Point'

Byte and Switch: But from the way you're talking about this, it sounds as if the value proposition has changed a bit now that you're part of Sun.

Napolitano: I think the messaging is a little different, because the channels are different. The underlying technologies are really identical. So just how we sell stuff as Sun is a little different than Pirus. Pirus could never have imagined selling a subsystem with disks. That wasn't what we built. So it changes how the product comes to market. But the fact is, it's a huge channel for us as Sun. We already manufacture these large-scale arrays. There's a lot of revenue – I can't give you numbers, but it's a large number –and to tie our data services platform on to that is a huge vehicle to enter the market. If you look at the different choices on how to go to market, it's simple math. It's staggering.Byte and Switch: OK, you've said the Pirus box is in the hands of a few beta customers. But looking at the arc of this technology, two years later the jury is still out on virtualization, right?

Napolitano: Well, yes and no. There are really two answers on that. One is the whole proposition of virtualization is really just being refined in the industry, and we're trying to target it around solutions – which is why the first standalone product comes out via professional services... The idea of selling virtualization isn't really the point. You have to sell solutions, so that's what we're doing.

And then, if you look at our approach with the integrated product, where you take the Pirus technology and integrate it into a large-scale array, there you're beefing up the functional and performance and scaleability characteristics of a large-scale array. Once again, a tried-and-true proposition. This is about how to get to market... without trying to evangelize something new and different where people get nervous.

Byte and Switch: But the original idea of Pirus was to have the switch sitting in front of a JBOD [just a bunch of disks].

Napolitano: So we still do that.Byte and Switch: OK...

Napolitano: That's the standalone product, we call it the Data Services Platform, sitting in front of heterogeneous storage, from a number of different vendors. When we come to market, we'll actually have a proved vendor list that we've tested with already. And we'll continue to grow that list over time.

NEXT: Sun Spots

Byte and Switch: Why is it in Sun's interest to sell something that lets customers use whatever storage they want?

Napolitano: It's in Sun's interest to sell this because that's the end-user problem, right? The end user does have a heterogeneous storage problem. Sun is really about – I'm learning all this, and I've seen it now first-hand – we want to have end-to-end solutions: We want to sell you a server, we want to sell a data services platform, we want to sell you a storage array. But that doesn't mean we won't sell you pieces of that. It's kind of embedded in the culture. So, develop technologies around standards – in this case, Fibre Channel – and allow people to pick and choose which parts of the solution they want from Sun.In the end, people are more comfortable with a single-vendor solution. But the fact is that the world is not homogenous. People have Hitachi Ltd. [NYSE: HIT; Paris: PHA] arrays or LSI Logic Storage Systems Inc. arrays out there already. We think it's important for us to support those. Frankly, we're going to support some of those basic primitives even in the integrated product, because we think that's important for that as well. You'll be able to support other people's arrays behind our integrated product.

Byte and Switch: Overall, Sun is more of a server company than a storage company, right?

Napolitano: Well, server companies have always been storage companies. I started my career at Digital, which was a server company. But the fact of the matter was, a large percentage of the revenue came from storage, as it does at Sun. They tend not to talk about it because they lead with the server proposition, because they want to sell the end-to-end solution. Mark Canepa, my boss, runs storage. He's a Sun employee; he's not the storage EVP first and foremost. So the end-to-end solution is the biggest proposition in the whole company. You don't hear us talk about storage in the way you would an independent storage company. We just won't do that. We're not going to do that. We're not going to unbundle our numbers. But the fact of the matter is, it's a huge market for storage products. Sun is a very large storage supplier.

Byte and Switch: But in terms of company-wide strategy, things tend to flow from the server side, right? Isn't that a challenge?

Napolitano: No, not really. I think... Canepa was a server guy; he's very strong in the organization and is the champion for storage products. It's clear we can grow the storage business at Sun substantially. So we're investing in it. The action around Pirus and other things that are going on, I think, should be testament to the fact that we are serious about the storage business. And, frankly, I voted with my feet. I didn't have to do the deal. I believe we can really grow the storage business here.NEXT: Owning the IP

Byte and Switch: Among the virtualization switch players, there are two camps. On one side, you have Rhapsody [acquired by Brocade Communications Systems Inc. (Nasdaq: BRCD)] and MaXXan Systems Inc., which have "open" platforms that can supposedly run any software. Then on the other side you have companies like Pirus with fully integrated, purpose-built switches. Why is the Pirus approach better?

Napolitano: So, I love the "open platform" positioning, because what it says from the startups is that they couldn't figure out how to do the software – they could only figure out how to do the chips. That's really what it says. Frankly, many of the guys who came after us really tried to use that, but at the end of the day they don't have a whole solution. This is a system, and it needs to be designed together. And there are certain parts of it that you really need to own the IP [intellectual property]. Frankly, this is one of the reasons Sun acquired Pirus versus the 10 other guys they looked at. They decided, "We need to own the IP." That was a piece of it.

I think implicit in your question is something else... There's this notion that you have core switching – which is about fan-out and port cost – and then you have what we would call data services at the edge. Edge data services are more computationally intense functions, like a file system. And a file system is millions of lines of codes. At the end of the day the best place to run a file system is on a CPU... Those are not things you really want to port to ASICs.

Byte and Switch: So you sit at the edge.Napolitano: So we sit at the edge. Pirus's historical message was, we don't do Fibre Channel switching. If you want a Fibre Channel switch, go to McData Corp. [Nasdaq: MCDTA] or Brocade, or whoever. We have some number of ports, but it's not our intention to compete with those guys. That is still true. Now, having said that, if you think of us as the edge data services platform, then we want to run data services. So what you'll see is, certain applications we need to own because they're tied to the architecture, they're tied to the data path, they're mission-critical. They have to run at the speed of light. Then there are other data services, which are further away – we will open up our platform to allow some of that to happen.

Byte and Switch: But someone might look at that and say, "Sun is delivering this as a tightly integrated hardware platform, because then the cost of moving off that platform becomes higher." Right?

Napolitano: No, actually. If you look at the cost of the infrastructure, it's not around the switching. It's not around the stuff in the middle. It's around the back-end storage. So by opening up to heterogeneous storage – which is our religion – we give people those degrees of freedom.

Byte and Switch: When we talked with EMC Corp. [NYSE: EMC] CEO Joe Tucci recently, he said it's "insane" to think any substantial amount of storage intelligence will move into the network – there's some on the host, some in the network, and there will always be some in the storage array. Do you agree with that view? [See our interview with Joseph Tucci, President and CEO, EMC.]

Napolitano: He's right. I would agree with that. You know, we tend to be so black and white in this world. Especially in technology. At the end of the day, RAID today lives in the host, in the controller, and in the array. It lives in all three of those places today. And there are big businesses in all of those places...Byte and Switch: But moving intelligence to the edge changes the locus of control in the storage infrastructure, right?

Napolitano: I don't disagree, and in some ways it's not in Tucci's best interests for it to happen... The point is, it was a great business model for [EMC]. And they made a lot of money. And life was good. And they had a great solution. But what we see rolling forward is that people want that degree of freedom to pick their back-end storage subsystem.

Byte and Switch: So the idea is the Pirus technology will be used as a way to get into data centers where Sun storage doesn't currently play?

Napolitano: It's really multiple. That's one proposition – which is our support of heterogeneous storage allows us to sell into accounts without necessarily having to sell our storage. That's true. But what we've found is people want to buy our storage, but they want to be able to keep their old stuff working. Storage is very different from networking components... People don't throw anything out.

NEXT: iSCSI? Oh Yeah, ThatByte and Switch: Pirus was big on iSCSI. Sun is not a big iSCSI supporter. What's the status of Pirus's iSCSI technology? What will happen to that? [See Sun Says iSCSI May Be a 'Mistake'.]

Napolitano: We're certainly focused on getting to market in a way that generates the most revenue for Sun... So that's Fibre Channel, snapshot, dynamic multipathing, and adding other data services on to that. We will turn back on file system stuff in the future, and we'll turn iSCSI back on – when the market is there. I just don't see the market being there. I think the technology adoption of iSCSI is far slower than all the zealots have been purporting for a long time. The proposition of us being protocol- and wire-agnostic is still there... We're not saying, "Death to iSCSI." We're just waiting for the market to develop.

Byte and Switch: So where on the roadmap is iSCSI?

Napolitano: The first place you'll see iSCSI for us is in site-to-site replication. That will be the first incarnation. We're actually trying to get that out this year. But the schedule's tight, I just don't know. There's some ASIC stuff going on. First quarter next year is probably more conservative. But it's not intended to be local to the data center – it will be between data centers.

Byte and Switch: Looking back at the history of Pirus, would you have done anything different? Do you have any regrets? You said you voted with your feet.Napolitano: Yeah, I voted with my feet. My team is very excited. We're working hard. The interesting thing is that being part of Sun now, our strategy for what was Pirus – now the data services platform strategy – is so core to the storage business. So it's kind of exciting. A lot of times when you're acquired, you're an outsider. The strategy that they'd been working on – now I see why they acquired us – we fit so well where they want to go. We allow them to manifest the strategy much more quickly in real products.

As far as Pirus the company... I don't know. I mean, we led the pack. We were the first guys out there preaching intelligence in the infrastructure and [being] protocol- and wire-agnostic. I think we overestimated ourselves, how fast iSCSI would be adopted. So we probably spent more time on that than we should have. Because the market just wasn't there for that. That's one area... Other than pure execution things and people stuff, the reason why startups fail is they overshoot the market by too much. They overshoot by so much that there's no market there when they need to get to market.

Byte and Switch: Is there another startup out there in your future?

Napolitano: If you ask my wife, it will be very, very clear. This is my third one. Adobe, Adaptec, and now Sun. You start doing the math of the probabilities, and it starts getting very, very small of hitting another one.

Byte and Switch: Close to zero?Napolitano: So close to zero it's almost not worth it. There's a lot here I can do. There's a great opportunity to build a great storage business here at Sun.

By the way, when is Byte and Switch going to put up its "Hall of Fame"? I still think Pirus should definitely get the top spot on that.

Byte and Switch: Oh, uh, right... That's still in the works, Rich.

— Todd Spangler, US Editor, Byte and Switch

SUBSCRIBE TO OUR NEWSLETTER
Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like


More Insights