In November, Cygnus Solutions was acquired by Red Hat (see story). Red Hat president Matthew Szulik described the merger as creating “a single, worldwide source that lets developers rapidly create Linux applications for servers and small devices — accelerating the adoption of open source technologies.” On January… 12, 2000, Red Hat announced that it has appointed Cygnus founder Michael Tiemann, well known for his strong embedded systems credentials and focus, to the highly prized post of Red Hat's Chief Technology Officer (see story). In this interview, we asked Michael Tiemann about the implications of the Red Hat / Cygnus merger on the embedded Linux market, its effect on the Cygnus EL/IX Embedded Linux API initiative (see article), and his general views on the growth of Linux in non-desktop embedded applications.

RL: How has Red Hat's acquisition of Cygnus impacted the priorities of projects you were previously working on at Cygnus?

Tiemann: Actually, the two companies are extremely complimentary, resulting in it feeling more like a reunification than an acquisition. One thing that's noticeable, is a pickup in the pace of activities. Open source software, by its nature, accelerates the rate of innovation, so when you combine Red Hat and Cygnus — two of the leaders in open source software — the pace naturally increases even more. The point is to out-innovate your competition. That's how you win.

RL: How do Cygnus and Red Hat “fit” together?

Tiemann: Cygnus opens doors for Red Hat that weren't particularly open to them before. Red Hat has been extremely successful in enterprise computing, whereas Cygnus' special expertise is embedded. What we're seeing today, is that the enterprise folks are becoming very interested in embedded — for all kinds of connected devices that aren't desktop PCs. So the embedded capability Cygnus brings to Red Hat is going to be enormously significant.

RL: How would you define “embedded”?

Tiemann: Any computer that's not a desktop PC.

RL: Does that mean you classify servers as embedded systems?

Tiemann: Maybe we should include “special function servers”, but not general purpose servers. There are lots of specialized server applications which have unique requirements that would qualify as embedded applications. Whenever you are compiling on a desktop system and need to download to a separate target system for testing, the application is likely to be one that qualifies as embedded.

RL: Linux has been highly successful in the server area, but much less successful as a desktop workstation OS. Could you comment on that?

Tiemann: A key reason for the success of Linux in the server market is attributable to its abundant open source protocols for TCP/IP, etc. In short, the answer to “Why is Linux so successful as a server platform?” is: “It's the protocols, stupid!”

RL: Why hasn't Linux been more successful in the desktop environment?

Tiemann: It depends who you ask. There are millions of developers today using Linux on the desktop–that's certainly a success. But if you want to compare Linux to Windows, no, we haven't reached that installed base yet. Improved ease of use (GNOME), better development tools (GNU), and more applications are coming to Linux every day. I don't think it's too far fetched to imagine Linux as a mainstream desktop operating system in the future.

RL: Let's talk about Cygnus' eCos operating system. How does eCos fit into the Cygnus / Red Hat picture?

Tiemann: In the non-desktop embedded market, there's a spectrum of computing requirements: from high end, to deeply embedded. The deeper you go, the more the embedded operating systems market becomes fragmented. There are literally hundreds of commercial RTOSes and thousands of roll-your-own solutions — all proprietary. The roll-your-own approach has been very prominent in the deeply embedded applications space, because it's usually necessary to tune the OS to the constraints of the application and its hardware environment. eCos provides the unique capability to be configured and customized — especially, because it's open source — so it is able to bridge the gap between the off-the-shelf and roll-your-own solutions.

RL: So is eCos still important after the acquisition?

Tiemann: Yes, we're still ramping up the eCos program. In fact, we have a long way to go. But already eCos is being deployed in office automation, digital imaging, consumer appliances, and other embedded applications. eCos can serve as a consolidating operating system for deeply embedded applications. Despite its growing popularity and tremendous flexibility, Linux isn't appropriate for the highly restricted environments of deeply embedded applications. There's still a need for “invisible” embedded computers — and many such systems can benefit from a deeply embeddable open source OS like eCos.

RL: How does the EL/IX initiative fit into this picture?

Tiemann: What we did at Cygnus, was to take the OS configuration technology that we developed around eCos, and apply it to Linux. The result is a scalable, configurable Linux API that can be applied to embedded and real-time applications.

RL: Was the Cygnus EL/IX initiative for Linux mainly driven by a desire to increase demand for eCos, or by the real benefits it could bring to Linux?

Tiemann: The goal was not to take market share from Linux, but to create an open API for RTOS functions in the context of the open source Linux model.

RL: What's your understanding of what transpired at the Real Time Workshop recently held in Vienna, Austria?

Tiemann: We (Cygnus) approached that meeting as an opportunity to share our EL/IX work with the open source community, for the benefit of the community at large. We presented what we had done, the technical issues and opportunities as we perceived them, and what we believe are the potential benefits of the EL/IX API to Linux. Our presentation was well received. The consensus of the group was to begin forward in an effort to use EL/IX as an open embedded/real-time API for Linux.

RL: Isn't EL/IX still a Cygnus internal project? How will EL/IX become a truly open, open source project? Who will maintain EL/IX?

Tiemann: We expect hundreds of developers to collaborate on the EL/IX project, just as has been the case with other open source projects that Cygnus has played a leadership role in (e.g., the GNU C compiler). However, Cygnus (Red Hat) will continue to be the steward of EL/IX API technology until it is ready to be placed in the hands of the developer community.

RL: When is that expected to occur?

Tiemann: “Real Soon Now ™”. Already we're collaborating with key technical members of the Linux community.

RL: How does EL/IX compare with existing real-time Linux implementations already in use, such as RTLinux, RTAI, uCLinux, etc?

Tiemann: Basically, the developers of all of these real-time Linux implementations are in agreement, technically, as to what needs to be done. However, it's quite natural that there will be some resistance to change, because these various approaches are developed and deployed in many systems. So, their developers have a vested interest in keeping them substantially the way they are. This same situation has traditionally prevented standards from being adopted in the RTOS market. But today, one thing is very clear: 80-90% of the real-time system developer community is desperate for an open standard real-time API. That's why the proposed EL/IX API for Linux will succeed.

RL: Thank you!

 
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.