NYCBSDCon Summary Part I

I arrived in New York City to unusually warm weather for October. I was glad that the weather was predicted to stay warm and sunny throughout the weekend of the conference, although I knew the list of fascinating talks would keep me inside most of the day. Not only would the conference be a great chance to learn about emerging technology, but also one of the few times that I get to talk face to face with some of the people that I have interviewed.

The Friday evening gathering at Havana Central was a great place to get started. My brother and I arrived a little late, but we found the tables of familiar t-shirts and grabbed some open seats. I had a beer and a Cuban dish, introduced myself to the people who could hear me over the din of the crowd, but didn't stay very long due to the early start of the conference in the morning.

Saturday morning came early, and with audio and video recording equipment in tow, my brother and I hit the dim and empty morning streets of Manhattan. Finding the right building on Columbia's campus was easy this year due to the number of posters that were visible on just about every corner, and we arrived with plenty of time to set up equipment and tape down cables. Registration was smooth, breakfast was plentiful, and the wireless network was stable. I was glad to see a vendor table marked for OpenBSD, as I usually buy the CDs at conferences. There were also plenty of tables offering t-shirts, pens, magazines, and books.

After the opening kick-off, Adrian Chadd gave an enlightening talk on high-throughput concurrent disk IO. I'm not much of a programmer, so I was glad that there was an overview of the simple and traditional way to read and write to disk, along with how it fails to scale. The talk quickly moved into a comparison of the different strategies that Squid and Varnish use to move a lot of data, and also how different operating systems can affect the implementation.

The next talk with Jason Wright (jason@) was a humorous and eye-opening description of the efforts that are required to work around bugs in hardware. After the talk, I started to wonder how any of my computers keep from tripping over these hardware bugs, and I was even more appreciative of the work that OpenBSD developers do to encourage hardware manufacturers to release documentation.

Matthew Dillon followed with a very technical presentation on the HAMMER file system in DragonFlyBSD. The talk was geared more toward developers than system administrators, so much of it was over my head. There were many questions, and it was obvious that there was a lot of interest in the room. Matthew Dillon's laptop was using HAMMER, so it was nice to see that he felt that it was mature enough to trust with his data.

Lunch was my time to stream live audio, and I wandered around to find anyone who didn't have a mouth full of food. I had time to speak with vendors and participants, pick up a little conference swag, check out the new iXsystems laptop, and stare wistfully out the windows at the nice weather.

After lunch, Anders Magnusson (ragge@) gave a history and current status of the Portable C Compiler. There were questions about various design choices, and people seemed very positive about the project. While pcc is known to be very small and fast, he noted that there was still some work to do on both the compiler and software with quirks specific to gcc, although it is available in the base install of both NetBSD and OpenBSD if people would like to try it.

Michael Shalayeff continued the theme with his talk on porting PCC. Many hardware platforms are already supported, including some really strange ones with 36-bit registers, but he emphasized that porting doesn't take much effort and people should dive right in and give it a try. During the end of the talk, both Anders and Michael were on stage to answer further questions.

There was a brief afternoon break, and then Julio Merino Vidal took the stage to talk about the Automated Testing Framework for NetBSD. The goal of the project is to replace the sometimes haphazard testing utilities that are available in NetBSD with a consistent set of easy to write and easy to automate tests. The project has also been designed to work on a variety of platforms, including Linux and OSX. The hope is that formal testing will become a greater part of the development process, and contribute to code quality across all platforms.

Jeremy Reed gave a timely presentation on implementing DNSSEC. With the recent publication of tools that exploit fundamental weaknesses in the DNS protocol, it has become necessary to use strong cryptography between our name servers. DNSSEC has been available for many years, and he demonstrated the fairly straightforward technical steps to set it up, but it quickly became apparent that there was a lot of work to do on the political front before the system could become pervasive on the Internet.

The first day of the conference ended with a funny and encouraging talk by Jason Dixon (jdixon@) about the relative merits of the BSD and GPL licenses. We were treated to a history of the licenses, some jokes and audience participation, and finally a reminder that nobody has a monopoly on the definition freedom.

The conference hall emptied quickly with the announcement of an open bar cocktail party sponsored by Sun Microsystems at Havana Central. There was a good turnout, although many people left earlier than I would have expected, perhaps due to the 8:45am kick-off in the morning. Knowing that I had to set up for audio recording in the morning, I followed the early crowd out of the bar.

Thanks to Will Backman for this great write-up of the first day of NYCBSDCon 2008, and to everyone who made the conference a success. Stay tuned for the next summary from Will.