News Archive (1999-2012) | 2013-current at LinuxGizmos | Current Tech News Portal |    About   

Executive Interview, Michel Gien and Michel Genard of Jaluna – Part III

Jan 21, 1997 — by LinuxDevices Staff — from the LinuxDevices Archive — 1 views

Part III — Jaluna's perspective on and predictions for the embedded Linux market

Q24. Can you characterize the state of embedded Linux, as a market and as a technology?

In terms of market segments, the one where we see a lot of Linux traction today is telecom devices and telecom infrastructure. People are moving to Linux for cost reasons. Traditionally, we've been very close to this market. We see very clearly that the industry is moving in this direction.

One of the roadblocks is the migration cost. Moving applications that people don't even know how they were written, you know, ten years ago by programmers who aren't there anymore. Moving applications to Linux is something they can only afford for new projects, and new products. A lot of the interest in Jaluna is that it allows people to bring Linux into their systems without disrupting what runs today.

Q25. You've also had some news about set-top boxes — I think you support a DSP from Texas Instruments. Are you involved in consumer electronics products, or are you still mainly focusing on telecom?

Within devices, we segment into two areas. One is the mobile phones. We see a lot of traction for embedded Linux in the mobile phones market, particularly in Asia. All the chipset manufacturers in the phone business are starting to have Linux offerings.

Another area is set-top boxes and customer premises equipment. In set-top boxes, there were two challenges. One was to have Linux running on the DSPs, and we've done that adaptation. The second one is to be able to simultaneously run the video codecs and Linux on the DSP, without requiring an additional processor in the design.

Here the value of Jaluna is to be able to use some of the leftover cycles of the DSP. Particularly in the case of TI DSPs, when you run video codecs for MPEG4, or Windows Media 9, then there are some cycles left that can be used to run simple Linux applications. Particularly those that manage the screen, the remote control, support for the IP networking stack, and email, and Internet browsing.

In the systems we built on the TI DSP, Linux handles the screen and the Internet interface, and the DSP-BIOS supports the video codecs, decoding video within Linux-managed windows.

Q26: So you're running OSware on the DSP, and through that running the DSP-BIOS and Linux?

A26, Michel Genard — Yes, side by side. It's another example like a hybrid car. So instead of trying to run the TI DSP BIOS inside a Linux thread, the DSP supports the codecs, middleware, and high-performance real-time features. And Linux shares the resources of the DSP to run other applications and everything that is not real-time.

Q27: Is that shipping in products today?

A27, Michel Genard — Hopefully we will be in a position to disclose a system soon that will go into production in the middle of next year. Customers are developing products with it, but not shipping yet.

Michel Gien — The TI DSP environment has its own development environment, DSP Express, that runs on Windows and lets users develop codecs around standard interfaces. In the case of the OSware-based systems, developers are still using their favorite development environment.

Q28: So you're really only penalizing them by using 100KB of memory resources, and the necessity to do the BSP for Linux?

A28, Michel Gien — What we did was to adapt uClinux, plus extensions to make sure it interfaces with the DSP BIOS. The 100KB is the low-level part for sharing the core, but then there are virtual drivers in there, and other APIs to build a complete solution.

Q29: Roughly how much overhead is needed for OSware in a small device like that?

A29, Michel Gien — It's difficult to say in general terms, because it depends on the performance of the codecs, and the number of applications on the Linux side. Normally, the way those systems are used, you don't browse the Internet or read email at the same time that you watch an IP-based DVD. You switch from one to the other. You do some control from Linux, and then some real-time decoding from the DSP-BIOS side. Really, this type of architecture is targeted at mid-range to low-end solutions, where you are ready to sacrifice a little bit of Linux performance to bring some of the Linux applications to low-end devices. On the high-end side, you would run two processors, one dedicated to Linux and one to the DSP.

Q30: What challenges do embedded Linux tools and OS vendors face?

A30, Michel Gien — The main one is the migration. A lot of people have forward-looking Linux based projects. They're evaluating Linux. People are sometimes not quite sure, beyond the fact that Linux is attractive, what's really in there, how they can really use it, what level of real-time performance is acceptable to the applications. To move to production systems, the migration costs of porting applications or redesigning applications is what we see as one of the biggest roadblocks right now.

Q31: What market opportunities do you see for Linux in the embedded devices and systems market?

A31, Michel Gien — Chipset manufacturers are a key part of the ecosystem that we're building around. They need to differentiate between each other, and provide software solutions with their chipsets, in particular as we move to complex chipsets with multiple cores, combined DSPs, graphics, audio, video — all kinds of combinations on the hardware side. Then they need to provide a software environment that allows their customers to readily use [their chips] in products. We see that as a big market opportunity for both embedded Linux and Jaluna, because Linux is becoming the standard OS that chipset manufacturers are going to provide to their customers.

Some customers will want other OSes, such as Windows CE, or Symbian. But the standard, generic OS that they'll provide as a way to validate the chipset and show stuff is going to be Linux. For us, it's an opportunity because we provide them with not only Linux support, but support for a choice of operating systems for the customer.

Q32: You're not partnering with chip vendors on Linux ports, though…?

A32, Michel Gien — It's not our business to do Linux ports for chip vendors. If we have a customer that has a chipset for which there's no Linux port, and they want to use OSware, we can use our skills to do the Linux port — we know how to do that. But this is not really our market focus. We'd rather partner with Linux vendors such as MontaVista.

Michel Genard — There's already too many Linux distributions. There is little value to bring here. The market needs consolidation. Hopefully, over time, there will be consolidation of an embedded Linux distribution as well. I believe that with Jaluna OSware, we are solving a problem that customers have today.

Michel Gien — You were asking about problems with embedded Linux penetration… one thing we see is the business model, and also some concern with regard to GPL constraints.

Customers in industrial and telecom markets have intellectual property that they are very protective about, because it's where they compete. Sometimes they have concerns about how the use of Linux is going to impact their IP protection. We see that as an opportunity for Jaluna, because we are able to separate the concerns and run the Linux applications in the GPL environment on one side, while keeping the proprietary software separated from the GPL Linux environment.

It's not a technical problem, but we see some decision-makers concerned with this, particularly the lawyers. Especially with regard to the FUD being developed around Linux, like the SCO suit. The big guys are sometimes concerned.

Q33: What embedded Linux technology developments do you find exciting?

A33, Michel Gien — The one we are really excited about, beyond what we've done so far with helping migration, is our security project. That, and the work being done within Linux itself to enhance security. You know, there are a couple of secure Linux or trusted Linux projects around. This is something interesting, because more embedded systems are becoming connected. The value of Linux is the openness, the availability of applications, the networking support — everything that makes the device open. If it's open, it's also open to diseases and contaminations and hackers and stuff like that. So, we see this as a big challenge, and a big opportunity for developments.

Michel Genard — People are thinking about multi-level security. What that means for a device, for example a consumer device, is that it's not a closed device anymore. It's designed at some point with one specific configuration of an operating system, with a specific Linux distribution, with a specific application. But by design, when it's deployed in the field, it has to accept a base new application, and maybe even a full distribution of Linux. So it's very likely that in the future, we will have a device that will include multiple instances of the operating system that have different levels of security. What's exciting is that all of this needs to be managed. You have to have some level of infrastructure that enables this, so that's very exciting.

Michel Gien — From a more general perspective, the application area is exciting. Sometimes we hear companies say they'd like to move to Linux, but compared to Windows, or Symbian, it's lacking applications and standards. A little bit like on the desktop, actually. It's important that Linux developments are not only concentrating on kernel technology and system software, but also on using Linux to support advanced applications and a large range of applications, such as user interfaces for smart devices.

Q34: Can you share one or two of your company's most exciting successes?

A34, Michel Gien — Our first success and validation, for me, is to have been able to raise $12 million from major investors based on our technology business. To be able to do that, you may imagine that the investors have been checking around, asking about the potential applicability of our technology. They talked with our early design wins to validate our approach from a business standpoint. There's not much we can talk about in detail, but that's obviously the first success.

Michel Genard — To give you some benchmark information about our investment… France is a small state, but anyway, Jaluna was ranked number four in terms of VC funding for 2004. That was in round one. It's pretty exciting.

Michel Gien — We have a couple of design wins in various areas. I mentioned network infrastructure, mobile phone designs with chipset vendors and phone manufacturers, and set-top boxes that we'll be announcing in a few weeks. We expect in the coming months to be able to disclose a lot of those successes. We'll do demos at the 3G show in Cannes in February on the phone side.

Being able to close this multi-million Euro contract with the French Department of Defense to build a secure system was also a big success, because it validates that our architecture is a good fit to provide security solutions, and we see that as important for the company in the future.

Q35: Did you say the mobile phone design win involves Linux?

A35, Michel Gien — Yes. You can understand that this is very, very competitive. The companies in that field all want to be first to market. So there's not much we can talk about there.

Q36: How about a failure?

A36, Michel Gien — So far, we are too young to have a failure. We wouldn't exist anymore if we had a big failure.

I can mention some failures with Chorus or with Sun. At Chorus, from a technical point of view, one of the mistakes was to try to do everything ourselves, including maintaining a full operating system, sometimes with several tracks. Now, we are focusing really on the area where we add value, and leveraging other people — OS vendors in particular — to actually provide the full solution.

From a business point of view, one mistake before was not to invest enough in marketing. That's why we got Michel [Genard] on board (story). But that goes back to Linux's challenges.

If you look at all the marketing that is done by Microsoft, compared to what's done on the Linux side… MontaVista's been doing a good job in making the Linux brand acceptable, but I think more effort needs to be made for the market to feel comfortable that this is something they can use.

Q37: Particularly in embedded systems, product vendors don't often tout the fact that their products are based on Linux, do they?

A37, Michel Gien — Sometimes we see people, especially integrators and service providers, who are scared to put “Linux inside” because there's no big brand behind it. At the same time, it's an advantage. It's bad for Linux because people don't know [that it's] inside, but the manufacturers can promote their own brand without having to promote Microsoft — they can promote themselves instead.

Michel Gien — Another thing that's interesting for me is that there are so many weird attitudes. People want to be religious and exclusive. Either it's Linux, and only Linux, or its all real-time, commercial operating systems. That's crazy. In this modern day, we are in a position to use a technology where technology is good, and to make different technologies work together, and make these things work smoothly at a reasonable cost.

Michel Gien — An operating system is only an enabler. It's not an end unto itself. The end is the product, what the user experiences. They don't experience Linux or any other OS. The OS is interesting for the OS vendor, but not for the customer. Linux is the best enabler around, especially because it's not branded, so you can enable your own brand on top of it, as a device manufacturer.

Q38: What's your vision of the future for embedded Linux? Will embedded Linux continue to have the kind of traction it currently has? And when will the interest in Linux really start to peak?

A38, Michel Gien — There's no doubt that embedded Linux will be dominant in the market. The timing issue is more difficult to predict.

Mobile phones have a lot of visibility, and Linux is going to be peaking there in the 2007, 2008 time frame. We'll start to see deployed 3G systems on a wide scale, and we'll see a lot of Linux-based devices there. That might provide the proof that Linux is workable in the device space.

We see a lot of new devices, prototypes of devices that are very weird, or in various configurations. But the mainstream is not there yet. I think the mainstream will be mobile phones.

Q39: How do you see the Eclipse platform fitting into embedded Linux development?

A39, Michel Gien — We think Eclipse is great. From an open source licensing point of view, it's a little bit more friendly than Linux. There's a little less problem with acceptance.

Q40: How about the 2.6 kernel?

A40, Michel Gien — Everybody's asking for 2.6. It's the latest version. When you start a new project, you want to go with the latest version. I'm not sure people really understand what's in it, compared to previous versions, and whether they really need the extensions.

From an architectural point of view, 2.6 is cleaner, and it's better to deal with from an OS management point of view. But I'm not sure that the features really make a big, big difference from a product manufacturer or user point of view. Maybe a little bit on the real-time side.

Q41: Thank you for sharing your thoughts with our readers, and good luck to you and Jaluna!

Story Navigation

Read all the interviews in our Executive Interview Series

Other Related Stories

 
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.

Related Posts:



Comments are closed.