Ohio LinuxFest

First, I missed the keynote due to a chair-to-keyboard error with my alarm clock. I did catch some of the comments by the keynote speaker, Paul Ferris of KeyCorp., but only the very end.

The sessions were set up so that there were always two going on at the same time. The first two choices were Apache Performance by Rich Bowen and Theory and Practice: SELinux on Fedora by Colin Walters. I chose to attend Apache Performance, as I have no use for Fedora now that I've found Mepis, love of my life. And I have no idea what SELinux is anyway.

The Apache Performance session was great. The presentation was rich and actually really useful. I'll admit that there were quite a few things in the "Don't Do This" category that I always did out of convenience. I won't make that mistake again. One great thing I loved was that every classic mistake given, he would give you numbers to compare the right way and the wrong way so you could see how much you screwed up by doing it the wrong way. That is a great way to teach tech-minded people. Throw numbers at them! He also showed how to easily benchmark your own Apache changes, so you can see how each configuration change you make can help or hurt your performance.

The beginning of this session was plagued by technical issues and I commend Rich Bowen for having such a great sense of humor about it. Don't worry, the microphone problems didn't hurt the talk a bit.

But I have one suggestion: Learn LaTeX syntax. These handouts are unreadable in places. Other than that, great job.

After lunch was over, the Birds of a Feather sessions were scheduled to start at 1:15. I was most interested in the MySQL BoF with Jeremy Cole of MySQL AB. I had been out of the MySQL loop for a while, working with PostgreSQL at work, and was interesting in seeing some of the new features they had worked on over the past two years. Here are some things that happened when I was gone:

I can only say, "Finally". I've always seen MySQL as The Better Database but it was always missing extremely useful functionality, the reason why I can't use it at work. Once 5.0 is stable, I might be able to talk my employers into using it.

One thing he said kind of bothered me. Someone asked him about when would someone have to get a commercial license for MySQL. He said that anyone that uses it in a commercial setting, like a hosting company, needs to have a commercial license. Then he kind of blew it off and said that MySQL AB wasn't going to crack down on any hosting company and they would really only try to get licenses from companies that advertise they have MySQL, but they probably wouldn't do that either. Well, what about when MySQL AB gets in a money crunch and decides to get some money out of these hosting companies and other commercial sites that are using MySQL? If there's anything smallish companies don't like, it's uncertainty. This might actually scare me away from using MySQL except for their license page that seems to indicate that it's only bad if you're shipping your product and not just having it used over the network. Of course it also talks about drivers, which worries me again.

At the start of my next session, Open Source Diversity by Deryck Hodge, the web desginer for samba.org, I noticed that they were recording and/or broadcasting the sessions with Powerbooks and iSights. I don't know what became of this, are they saving them sonewhere or was it a live broadcast, and if anyone out there knows, drop me a line.

Anyway, back to Deryck Hodge and Open Source Diversity. This isn't diversity as in racial or gender but diversity as in technical. The talk was mostly a kind criticism of writers like Eric S. Raymond and Paul Graham who believe that all hackers are alike and how this mentality prevents other people, who are brilliant but in a different way, from getting into our culture. According to Deryck, culture enforces perception and thinking and the Linux culture is enforcing a very technical oriented thought model that isn't necessarily reflective of where we are today. This has the by-product of keeping intelligent but not necessarily technical people out of the Linux community. These people could be helping with things like UI design, web site maintenance and documentation, but instead feel excluded from the clic.

I think this talk had a lot of really good points, but I felt let down in the presentation. It certainly got me thinking about issues of inclusion and promoting problems that technical people might not be interested in as just as important in the project as everything else, but the talk felt mildly directionless with no real payoff in the end. I think he was talking to two audiences at the same time; the technical, that might need a wake up call about how their attitude affects other people and the non-technical, that want to be a part of this big open source mish-mash. I think if Deryck could boil his point down into a more concise speech and possibly add some suggestions or success stories, this would be one hell of a thing to listen to. I think his write up on his site about Avoiding Monoculture Mentality is a step in that direction. With all of that said, I'm glad I was there to listen.

Next was a presentation on the VoIP software package Asterisk. It was given by Greg Boehnlein of N2Net in Cleveland. I'm not very interested in VoIP, but this got me thinking about looking into it more.

Next was a talk about SuSE Linux by Brian Maseck of Novell, Inc. This was actually pretty interesting as my current employer is using SuSE as our main Linux distribution. Some of the new features include AutoYaST for auto installation and configuration of SuSE Linux, usermode Linux, that lets you run independent Linux VMs under SuSE (this is part of the 2.6 kernel, but SuSE will have nice tools to let you work with it apparently), and it's various certifications, like ISO 15408 certification and Carrier Grade certification, that really make it the Linux for serious business. The QA process for SuSE was also pretty impressive. Before Novell bought SuSE, SuSE was doing an okay job with QA but was still expecting hobbyists to do most of it. Now Novell is going all out with the testing. One of the big stats was that they run through, automated and manually, 1500 Use Cases on each SuSE Enterprise release on a massive bank of different types of hardware. Very impressive. SuSE seems to be setting itself up as the Enterprise Linux solution while Red Hat is staying on the server. It will be interesting to see how this plays out in a year or two.

Novell will also have SuSE Certification, starting in the first quarter of 2005. It would be good to get some people certified at my company and get our product YES certified in the process.

The last session was Migrating to Apache 2.0 by, again, Rich Bowen. Another great session on what's up with the long forgotten Apache 2.0, released in 2002 to no fanfare. The first part of the talk was used to dispel myths about the stability and functionality of Apache 2.0 and the second half was used to show the limits in the stability and functionality of Apache 2.0. Yes, that was meant to be funny.

A lot of the MPM functionality was really amazing and state of the art kind of stuff. The idea of having a dedicated child process for each virtual host is pretty darn cool.

The main gist of the talk was, if you use any non-standard module, and this includes such standard non-standard modules as mod_php and mod_perl, then don't use 2.0. My need of mod_perl precludes me from even trying this version, but I can't wait until it's ready.

One thing he said was that the reason the old mod_s don't just work in the new release is that the API totally changed in 2.0. On purpose. So people would have to rewrite their code. Who the hell thought that was a good idea? If you want people to migrate, you had best get the most popular modules migrated too. I would say more than half the people that use Apache use mod_perl, so no mod_perl, no Apache 2.0, not a big surprise. Anyway, mod_perl for 2.0 is in a "any day now" mode so I'll look at upgrading then.

The event ended with a raffle held by Jon "Maddog" Hall himself. He was a very interesting closing remarks speaker and seemed genuinely excited about the grass roots feel of the conference. I would say that I agree with him 100%.

Some suggestions, put breaks between the sessions. The way the sessions were set up, there was no break between the end of one session and the start of another. You have to have buffer room in here for setup and teardown, something like 5 minutes would have helped a lot. Post details before the conference so people can plan ahead of time. I know that you can't plan perfectly before the conference starts and stuff pops up all the time, but it's nice if I can get an idea of what's going to happen rather than having no idea at all. Just post a tentative schedule online a couple of days before the conference. That's all I ask. Also, it wasn't really a LinuxFest. The Open Source world has gotten so big that calling this a LinuxFest limits us too much. Most of the talks had nothing to do with Linux, but with Apache or Asterisk or even FreeBSD. With Open Source. In the future meeting, this will probably be even more true. I think trying to point this out, that we're all in this Open Source boat together, would go far in building the community stronger.

Otherwise the speakers were great, the venue was nice and I hope to help out more next year. Great show guys.

And no, I didn't win anything at the raffle.

Update: Pictures from the conference! Thanks Skippy!