On the Record: Scott McNealy

...Q: Can you talk a little bit about Sun's relationship with Linux? How that is going?

A: We believe in community development. I'm going to sound a little Al Gore-ish here. But we invented community development at Sun.

The irony that he would choose that particular comparison is so wonderly rich it made my day.

The date that this entry contains does not properly reflect the reality of my situtation. It's the middle of the night - probably around 3am - where I am. I'm sitting in the lobby of the Hotel Kuebler in Karlsruhe, Germany because I can't sleep. It's very warm here and I'm stumbling over a distinct cultural difference between the US and Germany. In the US, my hotel room would be freezing. But it appears, at least in this part of Germany, that air conditioning is the exception, not the rule. I sympathize with this. When at home I wait as long as is humanly possible to put air conditioners in the window. But I often find myself giving in when I'm faced with trying to sleep in muggy upper-80s temperatures, and that's what we have here. So I'm craving air conditioning.

The first day of LinuxTag here in germany was interesting. It's a long show, going all the way until Saturday. I thought that I would spend my time being a booth babe for the full length of the show but so far I've spent the bulk of my time talking to people about Directory Server. There's a huge amount of interest here. People have been coming into the Red Hat/Fedora booth through the day looking for answers to questions. I also spent a couple of hours talking with partners who are in the booth who wanted to know more. It's been very interesting to discover how much the market has been waiting for something like this. (When I say "market" here I mean in both the economic sense and also as the "market" of free software components.)

On the last two days of the conference we're holding FUDCon II (The Fedora Users and Developers Conference) across from the show in the conference center. We're going to have a good number of talks on a lot of interesting subjects so I encourage that if you're in the area that you make the trip to come to the conference. I'm giving the last talk of the conference on directory server and intend to cover a lot of information about the project setup, what we're trying to accomplish as well as some of our future plans.

By the way, where are Novell/SuSE, anyway? This is germany, right? They have almost no presence at this show. It's very strange and at the same time very annoying because I expected the chance to see and catch up with some of my uber-cool Novell friends. At least I always have the gnome and x.org booths to visit and they are full of friendly people.

Ok, time to try and go back to sleep.

Today Red Hat is pleased to announce the creation of the Fedora Directory Server project. Our vision is simple: to build a free software project around the directory server technology that Red Hat acquired from AOL late last year. This software is pretty amazing. It's been the technology backstop for some huge installations that had redundancy and reliability requirements that are rarely found in the real world and it has proven that it is up to the task. It's fast, it's reliable, it's enterprise class, and now it's free software.

It's our hope that through this software release we'll not only be laying down the first layer for free software-based identity management solutions, but also creating a competitive advantage for all free software projects because they will be able to integrate with this excellent piece of software. This project represents a huge opportunity for everyone in the free software world and we're happy to have the chance to make that possible.

The term "identity management" is thrown around a lot in the industry and press these days. We define it as the secure control of user information and access rights across multiple business contexts - in short - ensuring that the right users get access to the right systems, data, and applications quickly and securely. You can't do that without a centralized service, and in the free software world we envision this as the place to store that information. This technology competes with the larger directory servers out there - Microsoft's Active Directory, Sun's and Novell's Directory Server technology. In and of itself, this software does not represent all of the technology that's required to compete in the identity management space but it is arguably the most important piece.

We have some pretty clear objectives with this project. We want to make regular releases of the software that work in the real world and get testing. We want to draw developers into the project, make them stakeholders and full contributors. They need to feel that they are equals. In this sense, we will be community-based open source. Red Hat may be starting this project, but we hope that we will be able to bring in a large amount of non-Red Hat participation as well. We believe that this is an important key to its long term success. And not just as part of Red Hat's software subscription offerings, but also as the starting point for identity management solutions.

One of our larger technical objectives - as I've said - is to integrate with as much software as possible. This means that when possible we're a configuration store for every application on a system. Every user pref. Every service on your machine can store its configuration in one of these servers. Have you ever had the vision of dropping a machine on a network and having it come up, self-install, and just start working? We'd like to see it too because it offers compelling cost of ownership argument that we think free software is in a unique position to provide. But it requires participation from the larger software development community. This means you and your project. We need to start working more closely together if we're going to make it to the next level of integration required to compete. At Red Hat we think that this is an important technology that is an enabler of that kind of group participation. We hope that it will be.

If we're talking about where we want to be down the road, we should talk about where we are today. Our strengths include excellent client side tools. There's the LDAP SDK that's been part of the Mozilla project for years and has been lovingly maintained. We've got Java bindings for the SDK. And we've also got the widely deployed ldap client libraries that have been part of openldap. Lots of different projects out there already use the openldap libraries to connect to ldap servers.

The server itself is in really good shape. It's got password-sync with NT Active Directory Servers, Multi-Master replication, replication over relatively slow WAN connections, reliability you can depend on, amazing scalability and it's even sports excellent performance.

On the weakness side, we've got some work to do in the short term. There are a lot of smaller problems - annoyances, really. It's really hard to build the directory server. Unlike most free software projects we have yet to move to using autotools throughout the build. This is mostly a result of history. In the closed source world there is little difference between making a build and making a release. So the current scripts do both. It's also the Netscape heritage poking through. The build is largely built on coreconf, which many readers will remember as the config and build system that was used as part of the original Mozilla software release.

In the past, installation of the directory server was done interactively. On most Linux distributions it's assumed that you are always be able to do a non-interactive installation of the software and get something that works out of the box, even if it's not completely configured. Right now our installation instructions reflect the fact that we're not very good at that. We need to make improvements to fit into the model that most free software hackers and users expect.

One of our larger problems will revolve around the set of decent Java-based utilities that we have for managing the directory server. These components are not yet open source. If you download the binaries that we have available on our pages you will discover that they include the free software directory server and these administration utilities. They make administering the directory server much easier. And it's our intention to open source them as soon as it's possible. But this begs the larger question: what do we want to do with these administration utilities over the long term? They don't work very well on the free software Java stack and depending on the very proprietary Java VM from Sun/IBM/others is not a very attractive option in the free software world. So our challenge is to build a long term strategy that continues with the legacy of these good admin utilities but allows us to build and run on free software. It also might be as simple as porting these utilities over to a free Java software stack, but we haven't spent time on that yet. It's our intention to do so.

We're left with a monumental challenge -- one that we see as being the challenge of the entire free software community. How do we start to find those integration points with the rest of the free software world and help to deliver a compelling and well-supported identity-driven software solution? We know we can't do it entirely on our own. As an example, think about the fact that if Samba 4 were combined with a Kerberos implementation and backed by this directory server we would have a high-quality, high performance, interoperable Active Directory / Novell Netware replacement. We have most of the hard pieces required to make this happen; I think that for the most part the problems to bring about something like this would be more political than technical. Bringing together these disparate communities is the challenge.

The same could be said of the desktop. Using the GNOME desktop as an example, how do we move forward to think about the desktop as part of an identity-driven network client? How do we start building a technical architecture that allows that kind of transparency? I think it's possible, but it requires a large amount of community participation. Mindset is as important here as anything else. Integration needs to be our one true creed.

We hope that by releasing the directory server as free software we're able to start the community down the right path. We hope that you'll help and join us.

Press releases are on the wire. One for the Fedora Project and one announcing support from Red Hat for the Red Hat Directory Server product. Sadly, the Fedora release points to the main project url instead of the the directory server url, but most people will find their way.

There are also a number of other stories that were written recently about this release. There was a slashdot post, an eWeek story, and an article on Microsoft Watch. It's a shame that most of them went to press with very little content but I guess that's what happens when you only have leaks to go on.