Foreword — This interview with embedded pioneer and MontaVista CEO Jim Ready explains why Ready is preparing his small, privately held company for the “battle of the century.” It reveals Ready as a big thinker willing to extemporaneously discuss today's computing trends within their historical context.

This interview took place April 21, 2004, as MontaVista was announcing a new round of financing. It is the eighth in LinuxDevices.com's ongoing series of interviews with influential executives in the embedded Linux industry.

---

Q1: How do you feel about having the combo of Wind River plus Red Hat as a competitor?

A1: Wind's capitulation — to the extent to which it has happened — I'll take. But, it's a distraction, independent of how successful you think Wind will be. You have to be prepared. They're not stupid folks, and you have to prepare for the consequences.

But we don't think they'll be that successful. It's simple to pick it apart. Which would you rather do? Everything twice, or everything once? As a pure [Linux] play, we get to do everything once. Wind has to do it twice. That's just a weaker position.

I don't think you need to have VxWorks and Linux to compete. Our business model is absolute proof of that.

They have a resource allocation problem that will never get better. You could equate Wind River to Digital Equipment (DEC), and Jerry Fiddler to Ken Olsen. That makes Fiddler's “No one's ever going to use Linux” analogous to Olsen's question “Who needs a personal computer at home?” DEC always had to split loyalties between VMS and VAXworks and Unix.

It doesn't mean Wind goes away, but they will struggle.

Q2: The Wind River/Red Hat partnership, along with Novell's acquisition of SuSE (which has expressed some interest in embedded), suggest a growing use of Linux as an end-to-end solution spanning the full range of computer systems and devices, from enterprise to embedded. How will MontaVista play in that market, which includes mobile devices, infrastructure systems, enterprise systems, and web services? Perhaps through a tighter partnership with IBM?

A2: We're the guys with the dominant market share in the device space. We take it all the way up through the enterprise side, with Carrier Grade. One of the reasons we do well with IBM is that they share the same unified vision of Linux everywhere.

There are light bulbs going off for folks about what's possible. MontaVista is positioned to be a beneficiary of the realization that Linux is a great alternative to Microsoft.

Q3: Well, Wind River actually is the market share leader in embedded. So potentially, an alliance of Wind River and Red Hat combines to span the full gamut, from embedded to enterprise to the desktop, doesn't it? Additionally, isn't this end-to-end play something Microsoft is going after in a major way?

A3: It's clear that VxWorks is dead. There are vanishingly smaller new design wins that incorporate VxWorks. Our semiconductor friends in Asia say that in terms of new design wins, the choice is overwhelmingly Linux. That's the whole reason we got into this in the first place — the steam has run out of old embedded technology.

Whatever Wind does in Linux, they have to split with Red Hat. That weakens their position.

MontaVista didn't get to where it is in terms of Linux and the larger embedded market by having underfunded competition. Over 100 million bucks has been spent against us. There's no evidence that Wind will expend anywhere near 100 million to get where they have to go. I don't see where the investment's coming from that can alter the landscape. It's not a little money game, it's a big money game.

It boils down to two things: Microsoft has infinite money, and the rest of the world, combined around Linux, has infinite money. The governments of China, and Japan, are aligning interests around alternative interests. This is the battle of the century. We're at the center of it.

The economics say Microsoft spent $95 million dollars on CE/Wireless space. But the rest of the world probably spent at least that much around Linux. Precious few technologies have that kind of investment level.

Q4: Would you compare the emerging battle between Linux and Microsoft to the Cathedral and the Bazaar metaphor described by Eric S. Raymond?

A4: The standards organizations like the Open Source Development Labs (OSDL) with its Carrier Grade Linux are demonstrably shining stars, that bring a semi-cathedral approach to open source development. Its an aggregation of TEMs (telecommunications equipment manufacturers), and companies like Alcatel and Nokia, specifying and then delivering compliant software. It's an organized bazaar, with people acting in combined self-interest to Linux where it needs to go.

CELF (Consumer Electronics Linux Forum) is a similar attempt and activity, but it's earlier in the game.

So, it's not black and white, and we don't have extremes of the cathedral / bazaar model in the Eric. S. Raymond sense. There are aggregations in the bazaar space that are more focused.

Q5: Traditionally, MontaVista has stayed away from market-specific products, such as providing complete OS/middleware stacks for PDAs, smartphones, set-top boxes, home services gateways, etc. Will that change? If so, which vertical markets interest MontaVista most?

A5: The alternatives are not strictly vertical and horizontal. We can point out the dead bodies along the vertical path, but the alternative is not a haphazard horizontal path. Rather, it's a horizontal path with strong partnerships.

The world before PCs is where we are [in the embedded market]. There were once very vertical solutions — computers that only did word processing, essentially, like the Sinclair ZX 80, or only did CAD. There was a lot of that, very dedicated devices. Along came a standard platform — IBM PC hardware and DOS — and it wiped out specialized systems. Like Daisy had their own UNIX boxes and Unix software for CAD — they had to bail on that.

Things are going to horizontalize. Cell phones, as you get into smart phones, will be more like platforms. You will be able to customize it yourself. And, if you have a Microsoft phone, you'll get a BSOD (“blue screen of death”) as you hold it.

There's probably an implication that says even though MontaVista will not be building vertical solutions, there is the qualification, testing opportunity to aggregating your ecology to have more vertical solutions that are qualified by us.

Qualcomm's CFO (chief financial officer) points out that the average cell phone today has the computing power of a desktop five years ago. Thanks to Intel and others, the power in devices does enable a different way of looking at things. The lightbulb is going on that we may be on the threshhold of a pretty fundamental change in computing that could change the rules.

Q6: How would you characterize the current state of embedded Linux, both as a market and as a technology?

A6: As a technology, there are still the customers who are under-satisfied, in terms of the capabilities. The appetite is enormous for moving Linux forward and taking it to realms unthought of. Linux is a wonderful starting point, but there are still gaps to fill. And that is a wonderful thing. There are market models that say where you want to be is having customers that are under-satisfied.

The OSDL Carrier Grade Linux's 2.0 specification is coming, with even more requirements. CELF wants to do a bunch of things, and then some.

So, between technology improvements and injections, we're strapped. Our engineers are working real hard. Software's getting pulled out of the building by our customers. We haven't reached perfection, so that's really good.

They've finally seen the light in the U.S. telecom market — that's good news. Ready Systems did a significant percentage of our business in telecom, but MontaVista is just starting to get revenue from telecom.

So as a market, we're trying to figure out a respectable growth rate. And, we're not satisfied. We want to get growing even faster. We detect a very strong demand.

Q7: What key opportunities do you see for embedded Linux? Any killer apps in sight?

A7: The smartphone. This “platform” idea that could develop around smartphones could be the battle royale. From an overall business standpoint and semiconductor standpoint, you gotta win it. It's the kind of battle that tends to bring in a lot of resources.

The home is another one, the battle of digital TVs and STBs (set-top boxes) and all of that. That also has done well for us.

And, of course, military/aerospace.

It all takes a lot of coordination. There are governmental things going on, and the activities of very large companies. It's by no means easy, and by no means done.

One of our good semiconductor partners who we've done well with and had significant design wins with — I can't say who it is… but we have given this company a lot of attention, and gotten a lot from them. I just got a letter from them that said, “Sorry for ignoring you in the past. Now, you're going to become a focus.”

Actually, two very significant semiconductor companies have told us this. “If you thought you got attention from us before, watch out.”

Q8: What embedded Linux technology developments do you find exciting?

A8: We do span quite a range. There's more stuff and technology in Carrier Grade than you can shake a stick at. A real mountain of technology. And same is true in Consumer Edition.

We're not done in real-time, at all. We helped push Linux along the real-time dimension, and we have made progress, but you're never done. To the extent there's any gap in regular old Linux, the extent to which there are performance gaps relative to real-time performance, we've got our eyes set on eliminating those gaps completely.

I would expect in the not too distant future, to eliminate any significant differences between Linux and the classic RTOS (real-time OS) stuff I used to do in terms of real-time performance.

Because of hardware and other things, in general, there's a lessening requirement for the classic capabilities of real-time OSes. But, there can be exceptional or extremely visible situations in which a key application area turns out to have a specific real-time requirement. Fundamentally, we've decided that we could probably ignore all this, but it'll be fun to close whatever gaps remain. This will hasten the demise of proprietary real-time OSes. That's a personal interest, given my background. I don't mean to slight other technologies, such as Carrier Grade and Consumer. But having a strong real-time story is one of the stronger perceptual barriers to wider Linux adoption.

Q9: How do you see the 2.6 kernel affecting the embedded Linux market?

A9: Actually, there were more relevant incremental improvements from 2.2 to 2.4. We have already mitigated differences between 2.6 and 2.4. Having 2.4 stuff with real-time capabilities [backported from 2.6], softens the pressure to move to 2.6.

Still, we'll be shipping 2.6 this year. From a customer point of view, there's pressure. It's not overwhelming. We did a pretty quick 2.4 upgrade, but followed in six months with another one. It's somewhat arbitrary when 2.5 becomes 2.6, and there's still a shakeout period. Waiting for it to mature minimizes perturbation.

Q10: How do you see the Eclipse tools platform fitting into the future of embedded Linux development?

A10: We used to make the joke about how we're managing 30 million lines of source code. Now it's 44 million lines.

Eclipse is a “gift from God” in some sense. Being cross-hosted killed Microtec [Microtec Research merged with Ready Systems before its 1995 sale to Mentor Graphics]. It's extremely difficult and expensive — enormously expensive.

People wanted to host on Linux, but also Unix, and also Windows. We're staring at that saying “Oh no! Here it goes again.” Then we heard rumors about IBM doing something with Websphere. Sure enough, that turned into Eclipse. It is a fundamental goodness in the Universe, and also has the power of obsoleting Tornado [Wind River's VxWorks development tool] as well. There's the benefit of being the new pure play again. When we started to do an IDE, we were able to do so with enormous efficiency [because of Eclipse]. We're on the Eclipse board. I'm by far the worst board member. I salute the hardworking other board members.

Q11: What affect will the CELF, OSDL, and other standards bodies have on embedded Linux?

A11: MontaVista is a key contributor to both, and very active. We're very pleased with OSDL and Carrier Grade, and we have high hopes for CELF.

Q12: What challenges do embedded Linux OS, tools, and services vendors face?

A12: It's died down, but there's certainly the strategic FUD against Linux, and I think that's becoming less credible. We've crossed a good threshhold in the last year. There are many very powerful folks who don't like Linux very much. That's a significant factor. Of course, Linux has a lot of support, too.

We're largely over this, but historically, some customers have had a naive view of how a commercial company would operate around open source, how the economics were. They had a naive view of what it takes to produce software. They'd think, “Oh, we'll just download it ourself.” Now, from significant customers, there's enough realization of the complexity and challenges around Linux.

Going forward, there's challenges. One is ISV qualification and interoperability. Linux is remarkably uniform. People are getting used to the idea, and gaining confidence that you can target Linux, instead of targeting a specific Linux.

One of the strengths of Red Hat is it has a lot of ISVs — some relevant to our markets. I'd like to undermine that by having a sense out there that if it runs on Red Hat, it'll run on our platform, too. From the greater industry standpoint, having an official comfort level of general interoperability is good for Linux, because Microsoft stuff by definition interoperates.

Interoperability and ISV qualification need to be further addressed — not technically, but emotionally. The comfort level of ISVs about what they can support needs to broaden.

Q13: What challenges does embedded Linux face as a technology?

A13: Real-time, both real and psychological properties; what we do with OSDL, with general availability; the universe of operating system technology that, as a community, we have to cover. It's a lot of technology pieces moving forward. OSDL is a nice way to aggregate some stuff, CELF as well. We're doing some of it, and many others are contributing, too.

Q14. Any comments on the SCO mess?

A14: They've been sliding lately. Things are unraveling nicely. Baystar redeeming its investment, without saying why, for example. My emotional or impressionistic feeling that I get is, “It's unwinding very nicely.”

Look at Red Hat's market cap, and how well we're doing. Even with what SCO has been doing factored into the “risk thinking” of the world.

Q15: What do you consider some of MontaVista's most exciting successes?

A15: I'll give you one general, and one specific success.

Generally, we have a completely broad range of customers. We've proven it's a good across-the-board general technology solution, when properly configured. That's neat, to have called that 100 percent right some time ago. It's very personally satisfying to see that happen.

Specifically, you've gotta say the handset stuff we've gotten into has been the most surprising, because of the usual thinking that Linux is too big, too slow, not realtime enough.

Motorola has released a total of three MontaVista Linux phones. The ability to pop out three phones based on the same OS is making people in the industry say “Oh sh*t, how'd they do that?”

It was no big deal for us — with a lot of work from Motorola — to do things that historically were not possible. Now, we can hold those up as the proof point.

The classic thing to me, if you want to see the old and the new — Motorola phones also have a VRTX real-time OS [created by Hunter & Ready Systems] in there. The whole confluence of things is emblematic of the ubiquity of what Linux can do when pushed and prodded to fit the requirements of customers.

Q16: How about a failure?

A16: We were not as successful with our forays into Java. We've realigned on that. We've gone exclusively to partnering. The specialization of Java got application-specific. There are people in the market that have Javas that are a nice fit. Unlike Linux which can be the common OS, there's more specialization in the Java space. So having a common technology across the board wasn't going to work. Java couldn't follow Linux across as many applications as Linux could go into. That's no technical dig, but just a specialization issue. So, we went to our ecology space for Java.

Q17: Do you still have a partner relationship with IBM for their Java technology?

A17: The Java technology is back with IBM, and they are a very good partner, as are a few other companies. We went from being an engineering and direct seller to being a partner.

Q18: How many employees does MontaVista have?

A18: We have 165 direct employees, and 52 engineering types under outsource that are very important to us. And yes, it's true, they're in Moscow and in Beijing. And, it's been a real plus for MontaVista. Toiling away around the world are over 100 MontaVista engineers.

Q19: What's your vision of the future for embedded Linux? How big will the embedded Linux market get, and when will it start to reach that peak?

A19: It's gotta be tough for the market analysts. Clayton Christensen has said that disruptive technologies are tough or impossible to predict. They're almost always bigger than expected. They come out of nowhere, and change things in ways that are difficult to predict. It's clear to me that this market always should have been huge, even on the proprietary side. Even on the proprietary side, there always should have been bizillions and bizillions in it. But it's been so fragmented, historically, with lots of in-house development.

Some of the estimates were a couple billion in 2003 in terms of the commercial side, with the in-house side being a significant fraction of that.

It's an unstable situation. Linux won't go into the future highly fragmented. We'd be stuck targeting two or three versions, and could never get critical mass. The promise of Linux is the promise of consolidation. Companies like RTI and others can have an ability to do something once and leverage it — a virtuous cycle. Fragmentation gets shaken out of industry. Linux is remarkably more uniform than anything we've seen in the last 20 years.

Software costs money. A lot of money. Relative to Microtech and Ready Systems, the amount of software we can produce, and the functionality of it, with embedded Linux far outstrips what we were able to do in the proprietary days. Embedded Linux is an unstoppable phenomenon. The future is extremely bright, and that's all I can say.

---

About the interviewee — James Ready, with over 25 years of technical and entrepreneurial experience, is a recognized authority in the embedded systems and real-time software industry. Co-founder of Ready Systems, he pioneered the development of the first commercially viable, real-time operating system (RTOS) product — the VRTX real-time kernel. Ready Systems, founded in 1980, merged with Microtec Research in 1993, went public in 1994, and was acquired by Mentor Graphics in 1995. During this period, James served as Ready Systems's President, and as chief technical officer (CTO) at Microtec/Mentor. James founded MontaVista in 1999 to provide the Linux operating system to the embedded systems market, and to offer embedded-system expertise to the open source Linux community. Jim got his BA from University of Illinois at Urbana-Champaign in 1971 and his MA from the University of California, Berkeley in 1976.

---

 
This article was originally published on LinuxDevices.com and has been donated to the open source community by QuinStreet Inc. Please visit LinuxToday.com for up-to-date news and articles about Linux and open source.