January 16, 2007

Fun and sun down under: Day one at Linux.conf.au

Author: Joe 'Zonker' Brockmeier

SYDNEY -- It took more than 17 hours in planes and a trip through customs, but I've made the trek from Denver, Colorado, to Sydney, Australia, for Linux.conf.au (LCA) 2007. Already it looks like the trip was worthwhile.

Linux.conf.au (or "Linux.con f.au," as it says on our misprinted hats) is a roving conference held annually in different locations around Australia. This year, the conference returned to Sydney at the Kensington campus of the University of New South Wales, where the first Linux.conf.au was held.

My flight arrived Sunday morning, local time. After a short trip from the airport to UNSW, I got the key to my accommodations and settled in. Some attendees are staying at hotels near UNSW, but it seems most of us are staying at colleges on the UNSW campus. The organizers put me up in Shalom College, which is about a five-minute brisk walk (all uphill, it seems) to conference venue.

If you've ever lived in a college dorm, or been in one, you have a pretty good idea what the accommodations are like -- functional, but not exactly luxurious. Still, with all of the activities going on, staying in the room isn't high on the list of priorities.

The only complaint that I've heard so far is that, as of Monday night, some of the colleges (including mine) still have no Internet access. We were assured that wireless access would be up no later than Monday afternoon on Sunday, but no such luck so far.

LCA kickoff

Jeff Waugh, one of the LCA organizers and someone who is known for his work with GNOME and Ubuntu, kicked off the proceedings with a short talk about the history of the conference and offered advice for attendees to get through the conference healthy and sane.

The first two days of LCA 2007 are dedicated to miniconfs -- talks and presentations about specific areas of open source. Monday's miniconfs included tracks on Debian, GNOME, education, embedded devices, virtualization, MySQL, and research.

The talks are mostly offered in 40-minute blocks, though a few miniconfs feature longer sessions. Though some attendees who are involved with specific areas may choose to attend only one track, most of us were hitting sessions in several miniconfs.

The first session I went to was in the Debian track: a "State of the Project" address by Debian Project Leader (DPL) Anthony Towns. I was hoping to get an update on the status of the Debian Etch release, which was tentatively scheduled to be released on December 4 but has been delayed.

However, Towns' talk was mostly a retrospective of the Debian flamewars that have occurred during and immediately prior to his tenure as DPL. Towns talked about the controversy surrounding the decision to implement qualifications for architectures to be included in a release, the controversy over the GNU Free Documentation License (GFDL), the controversy over the distributor's license for Sun Java, the problems relating to the Firefox trademark, and the controversy over the Dunc-Tank project.

Towns' talk was entertaining, and there was a fair amount of back and forth with the audience, as befits any discussion with a room full of Debian users and developers, but it left me unsatisfied, as I was really wanting to hear something more forward-looking about Debian.

After Towns, former DPL Bdale Garbee, now CTO for open source and Linux for Hewlett-Packard and president of Software in the Public Interest (SPI), took the floor to talk about HP's relationship with Debian.

Garbee talked about HP's history with Debian, and how and why HP is interested in being part of Debian. I was concerned that this could be naught but a sales presentation, but Garbee's talk was mostly nuts-and-bolts about how a company like HP can be part of a free software community like Debian, and what advantages Debian offers to HP.

For example, Garbee pointed out that HP has no interest in trying to maintain its own Linux distribution -- but it does have a need to implement features in a distribution on a timescale that is not possible with commercial Linux distros. Garbee says that Debian gives HP a time to market advantage, and gave the example of Debian's support for Itanium 2 with the ZX1 chipset in Woody (3.0), before other distros and before the hardware was available.

Garbee also touched on Debian with HP Telco Extensions, which is in widespread use in the telecom industry. It's not clear where his statistic comes from, but Garbee says that later this year, about 30% of all mobile calls in the US will go through an HP ProLiant server running Debian at some point during the call.

After Garbee's talk, there was a break for lunch. I participated in LCA's "hallway track," and introduced myself to a few attendees and asked how they were finding the conference. The attendees at LCA are almost uniformly friendly and eager to talk about where they're from and what they're doing with open source.

At 11, the talks picked up again. I headed over to the virtualization miniconf to see what was stirring there. Jeff Dike, author and maintainer of User-Mode Linux (UML), spoke about UML, Kernel-based Virtual Machine (KVM), and hardware virtualization. It was an interesting talk, though geared more toward developers than users and administrators (such as myself). Dike talked about KVM development, and how it would allow UML to improve performance by avoiding the ptrace system call when guest hosts need to do certain things, such as use the gettimeofday() function.

I also sat in on Kuniyasu Suzaki's talk on OS Circulation environment "Trusted HTTP-FUSE Xenoppix". This is a system where users boot a PC using a live Xenoppix (a combination of Xen and Knoppix) CD running Xen, which then downloads guest systems over HTTP-FUSE CLOOP. HTTP-FUSE CLOOP is basically a loopback device mounted over HTTP. The reason that Suzaki and his colleagues are working over HTTP instead of some other protocol that might be more latency-resistant is that HTTP hosting services are cheap and plentiful, and the protocol is much easier than some others to work through firewalls.

Suzaki was a bit nervous as a speaker, but I found the talk interesting. He distributed CDs with Xenoppix after the talk (always a plus) and I'm looking forward to booting it when I have some spare time and a more robust net connection to try it out.

After Suzaki's talk, it was time for lunch. The LCA schedule is just about perfect for providing enough time for talks and still some time for breaks.

The last talk that I went to on Monday was in the research miniconf, by Pia Waugh, one of the LCA 2007 organizers and part-time research coordinator for the Australian Service for Knowledge on Open Source Software (ASK-OSS).

Waugh's talk was titled simply "ASK-OSS," so I dropped in to see what it was all about. She had a slightly truncated time slot, since the speaker prior to her went long. However, she managed to run through her information quickly without losing the audience.

Waugh gave the background on ASK-OSS and explained what services it offers. ASK-OSS provides advice to Australian educational and research institutions about the use of open source.

The talks wrapped up for the day at about 5:30, and there were no after-hours activities planned for attendees. However, the organizers did put on a major shindig for the LCA speakers, of which I am one. We were bused to the waterfront for a harbour dinner cruise that lasted for several hours, during which we were treated to a spectacular view of the Sydney harbour, excellent food, free drinks, and good conversation. Beware of free software geeks primed with liquor -- it can get a bit ribald!

LCA 2007 continues through Friday. If you're thinking of attending at the last minute, I understand there are a few slots for registration left. If you're not in the area, check back here for daily reports from the conference.


  • News
Click Here!