|[Home] [Credit Search] [Category Browser] [Staff Roll Call]||The LINUX.COM Article Archive|
|Originally Published: Monday, 6 August 2001||Author: S.A. Hayes|
|Published to: develop_articles/Development Articles||Page: 3/4 - [Printable]|
Linux.com Interview: Brian Behlendorf
Linux.com sits down with Brian Behlendorf, President of the Apache Software Foundation and CTO of CollabNet to talk about open source, the meaning of freedom, Richard Stallman and aliens from outside our solar system.
|CollabNet||<< Page 3 of 4 >>|
CollabNetLinux.com: Let's talk about CollabNet, your current project. You are currently CTO and founder of CollabNet, can you tell us about the company, its focus and roles?
Brian Behlendorf: Sure.
So we started out with this idea that there has got to be something to this "open source thing" that can be turned into a process, that could be understood as a science, and if we could make it into a science we could make it repeatable and then we could take it to the rest of the software industry and say "This works, lets use it".
So we had a bunch of different ideas, we were almost like a laboratory in the beginning, and one of those ideas - SourceXchange - we gave it a shot, and well there wasn't the environment there that was appropriate for it, although I still think it's a good idea < smiles> but you need money to make a good idea into an actual successful company.
There was another idea that we started that has turned out to be a lot more interesting, and that is our SourceCast product, and the underpinnings of that is Tigris, which you might have heard of, it's hosted at tigris.org. The main idea is a software development infrastructure. You can think of it - well your readers are probably going to be familiar with Sourceforge: very similar to that. We pull together all the basic tools for software development and create an environment for software developers to create software collaboratively. And it could be for open source projects, it could be for shared source projects. One of our more successful projects these days is a giant shared source project with HP and a whole bunch of their customers
Linux.com: Is Tigris open source?
Brian Behlendorf: Tigris is open source, you can pull down the code, you can run it yourself, the main value add is in the form of services on top, providing it as an ASP, as well as integrating it into third party tools, and other interesting stuff. Then, of course, training for companies on how to build collaborative development processes and all that kind of stuff. So, we want other people using our tools, today we use the common open source tools like CVS and Bugzilla. We are spending significant amounts of time building the next generation of those tools, so, for example, CVS is a great tool but it is also about a fourth generation hack and we're building a tool called Subversion which is based on webdav and based on Sun and some other very interesting properties that bring in security natively and it's a native client-server application unlike CVS which had that bolted on at the last minute.
There's another project we have underway called Scarab that is an open source project for a Java-based bug-tracking tool.
Our business model does rely on building open source software and engendering its use. We want people using these tools. We want to see Subversion become the code versioning standard out there because we know that just like Red Hat benefits from Linux being the standard we similarly want to benefit from being the integrator and the whole provider of the commercial half of those components.
Linux.com: You probably know more about, or as much about, managing open source development teams as anybody else on the planet. How does that factor into your work at CollabNet?
Brian Behlendorf: Well, we are trying to build a process. You've heard of extreme programming, you've heard of other methodologies. Open Source development shares a lot of traits with extreme programming but has differences as well. So, we've been focusing on distilling that down to a science, to where we can train other companies on it and have sessions with engineers from those companies where we teach them how to put aside the ego from time to time and share code.
Many engineers inside software firms think that sharing their code with others would just be the worst thing in the world. So, you have to work with them at times, hear what they are saying, and hear what their concerns are. Sometimes our advice might be to not take a certain project open source or to hold off and re-implement it until it's ready to remove some barrier to adoption by others.
A lot of it is just figuring out what are the attributes of successful open source projects and how do we replicate that? Are all of them just a matter of chance? Maybe they are, in which case we don't have much of a business < laughs>, but I think a healthy amount of them are a matter of focus and a matter of learning about it and communicating it and distilling it down into action.
Linux.com: Many people have spoken about many successful open source projects having a leader, and I just heard one of your customers - Sun - talk about 'community managers". How do they figure into CollabNet?
Brian Behlendorf: We have a community development team who work for companies who pay us for that service. We'll participate on projects, sometimes up front but a lot of times behind the scenes. For instance if we have a project we will work with the company to identify the engineer who has both a wide scope of knowledge of the whole project and has the interpersonal skills: can communicate with people, can effectively write. What we do is help identify those people, help train them, help them understand what the attributes are of being a good leader in the open source community, how to carry on discussions, how to contain flame wars, thing that we know are very critical.
These are things that we know Linus does very well because it's in his genetics or something; he picked it up from somewhere...
Linux.com: It's a European thing. They teach it in high school in Finland
Brian Behlendorf: Oh really? < laughs> They should, they should teach this kind of stuff in high school. It's important for the future. Yeah, so basically it's just identify the good stuff and train people on it.
Linux.com: OK. What do you think are the principle benefits of "open source development" versus more traditional collaborative behind-the-firewall products like Rational and stuff like that?
Brian Behlendorf: OK, so the main difference as I see is tools like Rational, tools like Merant and others were never designed for the scenario that we see in open source. Here you have very geographically widespread developers - so the tools have to understand that they work over a WAN natively, a wide area network. They also were not really designed for what you see quite a bit in open source communities where you have core developers but you also have peripheral developers.
So, for example, my core is in Apache so I follow a bunch of Apache lists, but I also follow the FreeBSD mailing list because I run FreeBSD on my laptop, and a lot of Java mailing lists because obviously we're involved there. So, I have this peripheral involvement in those communities. I rarely have a need to submit a patch or anything like that but I do like hearing what is going on there.
In the traditional software development environment you had very segmented groups of who rarely shared things outside of their groups.
You also had those tools encouraging a model where the group in Boise would have their own infrastructure, the group in Seattle would have their own, the group in Sacramento would have their own and they'd be very disconnected, and so we have been focusing a lot on how do you get all those groups of people using the same infrastructure and what are all the implications of something like that?
Linux.com: OK, well I know a lot of our readers will want to know the answer to this next question: Are you hiring?
Brian Behlendorf: < laughs> Unfortunately, I'd have to say - well there is a job's page on Collab.net's site, we might have a few positions here and there but, overall, we're not hiring as much as we used to. Actually I used to have that in my .sig, but I had to take it out because we're not hiring so much any more.
Linux.com: Times have changed.
Brian Behlendorf: Times have changed. We do have a pretty healthy engineering team. It's going to be a matter of when - when people start spending more money, and we get even more customers, then we might start hiring again in the future. But right now, things are tight.
|CollabNet||<< Page 3 of 4 >>|