Roundtable "Potential of OSS for ATM" - Jointly organized by CALIBRE and EUROCONTROL
Previous 7th December 2005, EUROCONTROL-EEC, Bretigny-Sur-Orge, France Next

Home

Intro

Programme

Synopsis

Abstracts

Participants

Debriefing

Glossary

Organisation

Short bios

Thanks

Follow-up

OSS in Secondary Software Sector,
Voice of Industry

by John O' Flaherty, CALIBRE

Overview

OSS is a commercial reality, like it or not. In any case, EUROCONTROL will have to deal with it.

In a feedback session during a conference last year, we gathered a list of positives, negatives, and emotional reactions to open source. The top half-dozen from a ranked list:

  1. People issues: culture, social
  2. Commercial realities
  3. Technology
  4. Community
  5. Product: quality, features
  6. Legal: despite all the discussion around IP, it's not really that much of a problem. You just have to deal with it. Putative legal uncertainties concerning OSS are mostly marketing material, just Fear, Uncertainty, and Doubt (FUD) from proprietary software houses to attempt to discredit OSS.

Now, for EUROCONTROL. We've been asking the question all day, "should EUROCONTROL go open source?" Maybe we should rephrase the question as "Why aren't we doing open source? Why should it be proprietary?" You're a publicly-funded institution, running in a sector specifically intended to promote collaboration.

Let's first look at the technical infrastructure. Positive aspects include transparency, since you can see the source and assess its quality. You gain increased control from this transparency, and you can start to articulate the standards, at the lowest level. Software reuse is the big payoff. Hand in hand with open source are open standards. Negative issues include reliability. Can you use this for safety-critical systems, particularly in the ATM domain? How do you find the solution you require, out of thousands of active or moribund open source projects?

The social infrastructure is probably where the positive difference would be greatest. This is where you have to ask the question why not open source? If EUROCONTROL stipulated that in the absence of compelling reasons not to open the source code, all code should be open source, the perspectives for cooperation would widen. For example, in the AdaCore case, a single stipulation in the base contract created the fertile open source environment where AdaCore flourished.

Shared knowledge. How would the member states react? Would openness at EUROCONTROL stimulate openness for member states? Shared R&D costs are another positive aspect.

Critical negative issues in social infrastructure are again community. It's not easy to create a community. It requires resources and effort. When I visited the OpenATC web site, the thing that struck me was their mention that they had no resources. They couldn't succeed. Building community requires time, money, and support. When the resources are there, people can meet and share. They want to be part of the fun of it. Then you get the really great programs.

You need open source skills, and open source skills are definitely different. You need to get those skills. Either by training or by hiring, you need to get the skills to work with the community.

A more ambiguous aspect is that open source is global. It's not just European. The US, China, India - everybody will be in on the act. Your members may have issues with that. To have a community, you need to build trust and like any trust is about taking and giving, and to make it fun for the people involved.

J-P. Florent: If you improve the ATC in China, it is for the benefit of the passengers and the airlines.

J. O'Flaherty: So it does fit your mission.

M. Bourgois: We have no restraint in co-operating with anybody, because eventually it will benefit to the passengers, the ultimate clients of the system.

B. von Erlach: We are mainly a European organization, but since the ATM is a worldwide system, the benefit is also worldwide.

M. Bourgois: We will not focus specifically on China. If they are benefits and they use or abuse it, that is good.

M. Michlmayr: You can also have a support model while you have the expertise.

R. Schreiner: We develop an Open Source security system, and 40% of the investment comes from China and India. They really do like OSS from the US and from Europe.

M. Bourgois: We do work for ATM not for airlines. Even Chinese airlines flying European airspace will bring benefits for the ANSPs in Europe. The game could change once they start to produce ATM software.

G. Dean: They are likely to do that in the future, given that China is rapidly expanding its airline fleet.

J. O'Flaherty: So China is a questionable issue, probably positive.

In terms of legal infrastructure, the biggest issue is: Who do I sue, and who is going to sue me? Legal aspects boil down to those two questions, and the answer is unclear. But you can't just be neutral; you're going to get OSS anyway. As the PHILIPS presentation illustrated, open software will enter the enterprise in any case, you've got to figure out how to deal with it.

On the positive side, given EUROCONTROL's role, there is a strong motivation to be open and collaborative, and it does foster open standards. On the negative side, there's the ownership mind-set. Patents, used as a competitive marketing tool, are a bad thing. Spurious patents, rather than patents themselves are the issue. The biggest problem on the legal side is the lack of understanding of the issues. Companies dealing with open source have support centres to advice developers on legal issues. Perhaps EUROCONTROL could provide that support for ATM developers. Bigger companies can afford it, but the little ones can't.

The corporate environment for EUROCONTROL with open source would be a more cooperative, more flexible environment, offering greater job satisfaction. To develop great software, get great developers. The open source meritocracy fosters excellence, attracts great developers, and most importantly, helps retain them. On the negative side, there is a resistance to change. The tension between open source meritocracy and business profit motive will probably be tilted toward meritocracy, given EUROCONTROL's role. But high-level expertise, at high decision making levels, is another problem. If a member state decision-maker protests that a proprietary firm in their country will shut down if you move to open source, you'll have to justify their job losses.

Finally, in all successful open source projects, you always find a person, a champion for the cause. The presence or absence of a driving force champion is essential. If you have the champion, and you support him, the community will follow.

What would open source bring to EUROCONTROL? It would allow smaller companies to enter the field. Submitting good code could result in service contracts. But this applies to Russian and Chinese firms as well. Risk remains present, but can be managed. In terms of time to market, the impact is unclear. You can't just rush safety-critical applications to market.

Business models are also unclear for open source. Proprietary companies may have an advantage here: they've years of experience; they know how to respond to and win tenders; they can and will lobby officials. New OSS based companies need to learn that. Probably the best solution is hybrid, a mixture of open source and proprietary software. EUROCONTROL is in a position to provide the leadership for such an environment.

How do you sustain the open source? Someone has to fund the champions and the community. This takes time and money. You must have a hybrid fallback position: the real world is not open source or proprietary, but rather a mix of both.

In terms of acceptance, open source could invigorate, offer an open, progressive position, and provide good public relations. A downside is that you could be rejected by the libre community, or by the customers. Current vendors probably have something to loose with open source.

A suggestion could be to use open source as a focal point, a direction, a mission to go for, as Kennedy said about going to the moon. EUROCONTROL should choose OSS. Not because it is free or easy, but because to succeed in it will create a more valuable industry, because that goal will serve to organize and measure the best of the ATM sector's energies and skills, because that challenge is one that EUROCONTROL is willing to accept, one it is not willing to postpone, and one which it intends to win.

Discussion

from 17'45" to 30'00" (12'15")

M. Bourgois: Before anyone leaves, I would like to thank the people that have been involved today, especially the speakers who took time to share their thoughts, the other participants who brought additional insight, Jean-Luc Hardy for the organization.

H. Lueders: There's a lot of knowledge out there. There is high need to look more closely to the skill requirements that was just mentioned. We should have a special event to bring the experts together and address these hard questions.

M. Bourgois: That's a very good recommendation to CALIBRE and CALIBRATION. By the way, I want to thank them especially as well for co-organizing this event.

J. O'Flaherty: We actually did a workshop at Genova in July on Open Source education. There are a number of Master programs that teach OSS, but not at the industry level.

J. S.  Swierstra: Thank you very much for organizing this. It was extremely useful for us to see the legal possibilities, we had the impression that this would have been quite constraining.

In our division, we are looking at the transition from where we are today, to the dream system of the future. This means a paradigm shift in the introduction of automation in ATC systems. There is very little automation in the system today, and the system in the future is supposed to do it all by itself.

It's a very small community. There are three main players in this community, but they are not here today, which is an interesting signal. To achieve a minimum level interoperability, we must more or less force commonality on the functionality and technical solutions that these main players will have to provide to the users, the ANSPs. Even if they don't voluntarily want to go to some kind of open source in our small community, if we want to create advances for ATM systems, we, our division in Brussels headquarters, are absolutely convinced that we have to do something here.

We have to come to some kind of reasonable cut between what is really profit-making for these companies, where their expertise lies, and what is key point for us, which is to assure that we have interoperability. We don't want to say, "everything we do must be open sourced," that's not the issue. But there are certain elements which drive the essence of interoperability not only between ground systems, but also between ground and air systems as well.

There's a big scope for us. As you indicated, we need some organizations or persons to drive this, but there is a lot of wind against this. The fact that the three main players are not here today is a good indication of this. We have been discussing this for several years now, and the individuals from industry we talk to try to convince us that this is a very bad idea. This will not stop us from saying, "that is your vision, but this is the way we must go."

M. Bourgois: I would like to put your comments about the behaviour of the industry in context. The middleware standardization problems do illustrate that it is not easy to get the industry to take up open approaches or shared standards. However, for today's event, we did not push them to come, because we had limited resources. The target here was to confront internal projects of EUROCONTROL with expertise in open source. That was the stage of this discussion, and one should not read too much into the absence of industry representatives in today's meeting. A few actually are present today, and there's no sign at all of boycott by industry. But we do know they're not generally sympathetic to an environment which could create more competition.

J. O'Flaherty: It's not really an either/or situation. It would probably be to the benefit of the three major players if the open source approach is adopted. The PHILIPS presentation illustrated that nicely: it's a reality. They've just got to deal with it.

Also, if you open up a community around projects, you can attract other people. And from a European point of view, open source gives an opportunity for brilliant academics to contribute to mainstream commercial software. The community could get a lot bigger, with other players than you would expect.

P. Johnson: Can I suggest that the ANSP cannot wait for suppliers, because open is generally for the benefits of customers to the expense of suppliers who could be against it.

J. S.  Swierstra: With the privatisation, the roles of clients and suppliers in ATM are perhaps less clear. The ANSPs are not clients only. Some of them are becoming extremely sensitive to IPR issues. It is a very strange situation: one the one side, the ANSPs are our boss, on the other side they are bidding on the contracts that we are putting out.

H. Lueders: The interoperability issue is every day on the agenda of the European Union, and not only in Brussels, but also at the level of teh Member States. CompTia is everywhere involved. At the moment, we wait for the new communication by the Commission on interoperability. In the first quarter of next year, there will be the launch of the interoperability centre. I could add and add to that list, to say that it is high on the agenda. And the same for the IPR issues perhaps even more important at a longer term.

J. Feller: Just one final word of encouragement. Earlier today, I was asking Franco about the existence of competitor for their core product, the GNAT. What I was hearing was that there are people out there who build derivative works and then sell there expertise, but there is no one competing with them on their core product. I do not see that as a negative or surprising thing, because the core asset of Adacore is a unique knowledge of their particular space.

Now, open source started in well defined horizontal spaces. What we see now is a slow emergence of vertical spaces, like open source ERPs for example. The barrier that prevents open source of penetrating in these vertical spaces is a deep knowledge of the vertical spaces.

So, for EUROCONTROL, the one asset that you bring to the table is that unique knowledge. The same is true with AdaCore. And that is the key encouragement: if you want open source to work, you are the only people that could make it work.

M. Bourgois: Thank you very much, and I would like to conclude with this excellent statement.

Salient points

 
John O'Flaherty
 

"A suggestion could be to use open source as a focal point, a direction, a mission to go for, as Kennedy said about going to the moon."

Abstract article Slides MP3 OGG
abstract paper slides MP3 OGG