Article: An interview with Lineo CEO, Bryan Sparks
May 23, 2000 — by Rick Lehrbaum — from the LinuxDevices Archive — viewsIn this interview with Lineo CEO Bryan Sparks, LinuxDevices.com founder Rick Lehrbaum talks with Bryan about . . .
- Lineo's perspective on the wide range of embedded Linux solution strategies
- Lineo's approach to supporting the full spectrum of embedded applications
- Lineo's open-source philosophy
- Lineo's business and revenue models
- Lineo's corporate growth and the… integration of its many acquisitions
Sparks: It's especially fascinating that the proprietary RTOS vendors are adding Linux to their strategies. We first noticed LynuxWorks, and now QNX, doing this. We think this increasingly common migration to Linux by the traditional RTOS companies is in direct reaction to Lineo's embedded Linux offering, which we put in motion last July. But, we think OEMs will prefer our Linux-based embedded system software products and business model over the “hybrid OS” solutions offered by companies like LynuxWorks, QNX, Wind River Systems, and Red Hat.
RL: I notice you included Red Hat in the category of companies offering “hybrid OS” solutions. Could you elaborate on that?
Sparks: We include Red Hat in this category even though eCos, their low-end embedded product, isn't “proprietary.” It's open source, but it's certainly home grown. But it's not Linux. What they're doing is adding a Linux api on top of eCos, which isn't Linux. That's a “hybrid” strategy. They support Linux for the high end, and eCos for the low end. Lineo believes Linux is the superior embedded OS and that Linux is going to be much more appealing to OEMs than a hybrid solution, where they have to deal with two different technologies instead of the simplicity of keeping everything on Linux.
RL: Speaking of Red Hat, Michael Tiemann (CTO of Red Hat, Inc.) was recently quoted in EE Times as saying that Lineo appears to be buying up every company with an incompatible version of Linux — with the result that all the fragmented versions are now under one roof. Would you care to comment on that?
Sparks: Hey, at least he spelled our name right! The way we see it, Red Hat created EL/IX to justify their eCos strategy. We think that their basic premise is wrong. Tiemann says Linux is fragmented. We disagree! The Linux api stays pretty consistent, even when you move away from Intel processors. The only way Red Hat can get some support for EL/IX is to argue that Linux is fragmented. Otherwise, EL/IX is silly. We don't agree with their premise; we think the Linux api is pretty solid. Through the six companies we've acquired, Lineo has been able to extend the same (Linux) technology across multiple chip architectures, and we've also added real-time capabilities. Our acquisitions have given us a broader breadth of Linux support, from very small microcontrollers, through traditional platforms like x86, and up to high end, high availability systems. We see our strategy as a “breadth strategy”: moving a common api (i.e. the Linux api) in multiple directions, to gain broader market share.
RL: You speak of a “common api” — the Linux api. Interestingly, that's the very same expression QNX Software Systems is using in attempting to reposition their proprietary RTOS product as supportive of Linux. Do you think there's any truth to that, that the “Linux operating system” really consists of many more components than simply the Linux kernel? In other words, because they support the Linux api, they are supporting Linux?
Sparks: From the business side, you add a Linux api because Linux is where all the action is. All the third party vendors are developing their applications and tools for Linux. So, if you're a proprietary OS vendor, how do you take advantage of the Linux movement? But one of the big challenges that comes about, when they announce their support for Linux, is that they are questioning their historical business and revenue models. Linux puts per-unit price pressure on what they were able to get with their proprietary OSes.
RL: Whether open source or not, aren't there still a lot of different approaches to generating software revenue?
Sparks: Lineo also has a royalty based model. We add stuff on top, and around, and tools, and so forth. But one of the things that Linux does is put per-unit price pressure on what you can charge. For example, if you're getting $10/copy a copy for a proprietary OS, but you're advocating the api/abi of an OS that ships for a tenth of that, then that's going to put price pressure on your $10/copy product.
RL: I think what the RTOS companies are trying to do is provide an alternative kernel that supports the Linux api (or abi) but offers unique capabilities. Then, they try to sell the advantages of their kernel over Linux and if you want or need what they offer, you pay for it. The proprietary “Linux-like” kernel (QNX or LynxOS) fits in the same “socket” as Linux. So Linux has done something that it hadn't intended to do, which is to create an api/abi “standard” . . . which is actually good for the industry anyway.
Sparks: Absolutely! It's good for all those application developers and vendors. You write an application once, and you get all these platforms for free.
RL: Back to price pressure. You say Linux is putting price pressure on the traditional RTOSes. How do you feel about MontaVista's slogan of “100% royalty-free open source Linux.” Are they putting price pressure on Lineo?
Sparks: MontaVista keeps saying that a royalty-based approach is bad. We believe a royalty-based business model is the Linux model that will have the greatest long term revenue potential, while still respecting and supporting open-source and the rules of GPL. The customers we are talking to are looking for shortened time-to-market. They recognize that making sure we stay in business is good for them. Obviously, it's good for us too!
RL: I understand and appreciate that, from a business model perspective. However, it seems to me that the one weak aspect of that logic is that if you charge a royalty based on licensed proprietary technology, that seems to imply that it's not open-source software. That means the software you're charging a royalty for won't benefit from the whole evolutionary process that accrues from the open-source phenomenon.
Sparks: No, I don't agree with that. For example, we sell an embedded browser that we charge a royalty for. We sell a ton of licenses for that browser, in spite of the fact that Mozilla is free. We have found a tremendous amount interest in our browser for devices — handhelds, set-tops, and so on. We're able to charge a royalty per unit for our embedded browser, because it's uniquely customized for embedded applications. That has value to an OEM. The same thing applies to Linux: the fact that we move Linux to a Hitachi Super-H chip has value. The fact that our engineers do the work — even though we publish the results back under GPL — that has value. By no means are we stretching the GPL. If you want to work with us, then we can help you reduce your time-to-market and give you the product you need on the platform you want. The only area where we're going to be price pinched is on Intel, because that's where the Linux community does all their work. Fortunately for us, the bulk of embedded is not x86 based — it's based on other platforms. Our ability to charge a high per-unit royalty on x86 platforms is going to be limited by the fact that people can roll their own. But on the other platforms, even though people can roll their own, the amount of effort is much, much higher. So, the fact that we wrap it under a larger contract that provides maintenance, service, support, value added products, etc., doesn't diminish the fact that we comply with the GPL. We don't modify it at all.
RL: But your embedded browser is not GPL?
Sparks: That is correct.
RL: So, it sounds to me like you're making a similar argument to the ones that QNX and LynuxWorks are making, when they say they want to provide the option of alternate non-GPL, “Linux-like” kernels?
Sparks: Perhaps. But what they give up are the great inherent benefits that Linux provides, including things like high performance protocol stacks, support for peripherals, wireless support, etc. As a result, they're going to have to reproduce all of that themselves. I'm not saying that's a bad strategy — in fact, I think it's the only strategy they can adopt. Otherwise, Linux is just going to take over!
RL: In one way, it seems to me that the royalty strategy LynuxWorks has adopted is similar to Lineo's. Lineo offers customers a choice of a free open-source browser, Mozilla, or a royalty-based proprietary embedded browser, Embrowser. Similarly, LynuxWorks offers their customers a choice between an free, open-source kernel, Linux, or a royalty-based proprietary “Linux-like” RTOS, LynxOS.
Sparks: The interesting thing about LynuxWorks is that one of the things they do is a “bait and switch” approach. They say: “Linux is great, but if you want hard real-time, you need to use LynxOS.” We say: “We do hard — not soft — real-time with Linux.” There's no need to go to LynxOS. Another important distinction between the two strategies is that our proprietary component is the browser, which is an application, whereas theirs (LynxOS) is the OS itself.
RL: You mentioned earlier that Lineo now has a very broad span of Linux support — from microcontrollers to cluster computers. Red Hat has taken the approach of promoting eCos, rather than Linux, for very small devices at the low end of the spectrum. What are Lineo's specs for Linux at the low end, for microcontroller-like applications?
Sparks: We can do things like execute-in-place out of ROM, so our RAM requirements can be very, very small. And we also run on MMU-less devices. As a result, we can get full featured Linux — everything you'd expect from Linux — with very low system resource requirements. We also have the capabilities, through several of our acquisitions, to assist our customers in migrating from traditional embedded OSes to our embedded Linux.
RL: Could you provide some details on the low end requirements of Lineo's embedded Linux offerings?
Sparks:
- Rt-Control provides uClinux, a version of Linux for microcontrollers such as the Motorola 68k/ColdFire line, i960, ARM7, and ETRAX chips. Because these chips have no MMU (thus cannot provide multi-tasking capabilities), uClinux is able to run full-featured in as little as 150 KB of RAM with a 1 MB ROM chip.
- FirePlug focuses on specialized Linux-based project such as their Linux firewall. They build these on the ThinLinux product, which runs in as little as 2 MB of disk/flash storage and 8 MB RAM.
- Embedix, Lineo's flagship product, runs a complete multitasking, networked Linux operating
system in 2 MB of ROM/flash and 4 MB of RAM.
RL: Speaking of Lineo's acquisitions, I understand that Lineo now has over 200 employees, resulting from the combination of Lineo plus six acquired companies. It must be quite a challenge to get those disparate forces, resulting from your many acquisitions, moving together in a harmonious manner.
Sparks: Actually, it hasn't been a problem. Certainly, we've been logging a lot of miles on airplanes! But one of the key considerations prior to each acquisition is how well we're going to fit together, what the synergies are, etc. Obviously, we're doing integration. But we feel we've figured it out, and at this point it's mainly a matter of execution and rolling all those technologies into a single product roadmap. It's been really fascinating! We've got a lot of great people, and we've found that the process actually hasn't been as hard as we thought it might be. Incidentally, although it seems like a lot of companies to fit together, three of them — Zentropix, Moreton Bay, and RT-Control — were already working closely with each other as a sort of “partner trilogy” within the microprocessor and real-time space. I'd also like to point out that, through the acquisitions we've made, we now have local language, local time zone technical support for our OEMs in all three major markets — North America, Europe, and Asia. That was also taken into account during our acquisitions.
RL: Thanks very much!
Related stories:
Lineo registers for initial public offering
Lineo announces receipt of $37M funding
Lineo acquires ThinLinux specialist, FirePlug
Lineo acquires High Availability Linux specialist, INUP
Lineo adds Moreton Bay to its corporate portfolio
Lineo acquires USE, gains deeper micro support
Lineo acquires uClinux creator Rt-Control
Lineo Acquires Zentropix
Want to comment on this article? talkback here
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.