LinuxDevices.com Archive Index (1999-2012) | 2013-current at LinuxGizmos.com | About  

Demystifying Device Software Optimization

Jan 19, 2006 — by LinuxDevices Staff — from the LinuxDevices Archive

The “Device Software Optimization” (DSO) concept championed by Wind River has gained another proponent. In this brief interview, LinuxDevices.com speaks with Enea CEO Johan Wall about DSO as a potential antidote to exploding device software complexity, and the possible formation of a DSO industry group.

DSO was coined by Wind River's lead marketer, John Bruggeman, as a broad term encompassing that company's embedded software stacks and support services. It subsequently found limited favor as a broad industry term, primarily with other real-time operating system (RTOS) vendors, including mil/aero RTOS specialist Green Hills Software and now, with Enea. Among other embedded products, Enea makes an RTOS called OSE that it says can be found running the signaling functions in about half of all mobile phones.

Our interview below with Enea CEO Johan Wall aims to learn more about DSO as a concept, and about the potential for a DSO industry group. Enjoy . . . !


Q1. We hear rumblings of a DSO organization being formed. Can you tell us what its focus and charter would be, who might be involved, and what kind of an organization it would be?

A1. In my view, DSO is an industry initiative aimed at improving how device software is developed. It's part of the maturing of this industry, which is highly fragmented, and highly “technology-focused.”

I believe the key essence of developing software is to efficiently reduce time-to-market and reduce cost, while still producing excellent software. With increased software complexity, there has to be a better way. We're drowning in code, today. We see DSO as a way to bring the industry together to find new ways of developing software.

That's my view of DSO. Coming to the initiative itself… DSO was founded by [Wind River]. We have joined DSO because we see it as an important part of shaping this industry.

[The companies participating in the DSO initiative] are leaders in the industry, calling for all other players to get together to form some kind of trade body, or at least to outline where this industry is going in the future. We've been in talks with several of the players in the industry, and we invite everyone to join us. There are, at this time, no concrete plans yet for a Device Software Optimization industry body, but we are certainly contemplating that, along with industry tradeshows and other ways to spearhead this effort, because there has to be a better way of developing software.

Today, we're not reusing software to the extent that we would like to. There is a lack of interoperability between various software components. There's a lack of “layering,” as we see it. And thus there is a lack of achieving the clear benefits that our customers want to see.

These are some of the many aspects of how we develop in this industry that we can use DSO as a means of changing.

Q2. Sounds like there are a lot of different efforts that could be combined under a DSO umbrella. Some have been tried and failed, not once but numerous times, such as common APIs, driver standards, and so on. Eclipse seems to be bringing some integration and coherence on the tools side, and Wind River has started a device project under Eclipse. So it seems like this might represent a lot of different efforts that might be difficult to unify in a coherent way…

A2. I think you're making a very good point here. On the interface front, there's a need for technically oriented bodies to spec out interfaces. Eclipse is successful, but there's also a need that's more obvious now. Code complexity is increasing so quickly. There's missing systemic thinking here. We have to think and act differently.

We have to realize that the future is more about reusing software components, and delivering middleware and other components that are pre-integrated, and that perhaps contain a variety of software components. It might be Linux together with middleware from one vendor, together with a network stack from a third vendor, together with security software from a fourth vendor, etc. This is more than about tools — it's also about finding ways of reusing software components.

Q3. It seems that industry groups work best when they are organized around very specific tasks, or specific device categories. Can you identify any specific goals for a potential DSO organization to focus on?

A3. I appreciate your comments there. I'm sure there's validity to it. Our thinking is that this is an industry effort, rather than a technology effort. Or rather, an effort of a subset of the industry, part of maturing the industry, which has long been very technology-focused on solving specific important problems. Now, with software complexity growing so quickly and time-to-market requirements speeding up significantly, we believe there's a need for an industry-wide approach looking at how we reduce time-to-market, and how we reduce the bill-of-materials cost of the final product, while still having good quality software.

It is a broader approach than, for example, the LiPS Forum. But it is nevertheless very important. I have to stress that DSO is about freedom of choice. We want to include all the various software components. Given that Linux is becoming more and more important, Linux will play a central role in the discussions about DSO.

We see a level of code complexity that is just mind boggling. Enea is a leader in the field of RTOSes for cell phones. Some of our major customers might integrate 25 software components on a phone. That integration process and the effort to get all these components to work together is simply not sustainable, if we are going to keep development timeframes to a reasonable level.

Q4. What kinds of structured activities or technologies will result from DSO?

A4. Several things have been discussed among some of the industry leaders here. You could look at common trade shows around common themes, DSO certification, and at various ways of furthering the reuse of components to reduce development time. I'm sure we can come up with a list of concrete things. We've seen this in other industries. The Semiconductor forum, SEMI, has taken a similar approach, and successfully so.

Q5. Most early proponents of DSO appear to be RTOS vendors, such as Wind River, Green Hills, and Enea. Is there a place in the movement for Linux providers and users as well?

A5. There's nothing in DSO that is about safeguarding investments in legacy RTOSes. Enea provides an excellent hard RTOS, but we're looking to the future where the RTOSes will have to work together with the various middleware components, and other parts of the stack, and work together well. The age when everything was solidly built on a single proprietary RTOS with proprietary components all the way up to the application layer is long gone.

Q6. What would you like our readers to do? Is there a call-to-action?

A6. The call to action is to learn more about DSO, and take a hard look at how the development process can be simplified and accelerated. Look at what are the reusable components available to use, instead of writing middleware from scratch, for example, which some companies still do, amazingly. Those are some steps we'd like to see taken.

We have to look to the future to understand the importance of DSO. Code complexity and system complexity are increasing significantly, especially when we go to an open freedom-of-choice kind of world, where things have to work well together. We have to look toward the future. How will we manage that complexity in the future?

Q7. How are any efforts to establish a DSO organization progressing?

A7. The industry leaders we've been talking to and the contacts between us and various parties in the industry have been very well received. Of course, if there's an interest in DSO, feel welcome to call us, or get in touch, and we'll be happy to explain the benefits of the DSO movement.


About the interviewee — Johan Wall is President and CEO of Enea. He has a Masters of Science in Electrical Engineering from the Institute of Technology, Stockholm, and also studied as a visiting scholar at Stanford University. He previously served as president of Framfab AB, president at Netsolutions AB, and product developer at Verizon Laboratories, in Boston.


 
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.



Comments are closed.