Day 3 - New toys, short(er) presentations

Opening keynote/opening session

Dan has gotten quite informal with this and it goes pretty fast. The keynote started late (people still registering), so it was brief and just welcomed everyone, mentioned some evening activities (including the BSDcert beta tonight), what to after the conference on Sunday, and the fact that pgcon is on next week.

First session - Coverity

Code audit and testing software. Davis Maxwell gave this talk. While the coverity product is commercial, the company thinks that open source software would benefit from their tool as well, especially as they use open source software as well and ran across some interesting things in some libraries they use. David goes around to conferences talking about the tool and engaging open source software developers in trying the tool, hopefully using the tool. More info can be found in teh proceedings (I hope), and there was also an interesting story from Wietse Vename regarding the tool finding unexecuted code in postfix. If I was a programmer, I’d be looking at the tool. It looks like it will save you some serious runtime bugs.

Second Session - Open Source Security Lessons

A brief history of Wietse Venema’s software career from the early days of configuring email to tcp wrappers to The Coroner’s Toolkit to Postfix. A good talk covering the history of his career and the why and how of the tools he’s contributed to the community. There was also some interesting talk about how business ended up supporting open source. There are some lessons there that we could all use. The talk will eventually be available  through the BSDCan website (no details yet).

Third Session - Home Security / Monitoring with FreeBSD

Rob Krten gave a short talk on all of the interesting things you can do with an excess of computers, time and a desire to watch/log everything. Audio processing of the phone system, cameras watching for interesting things, weather monitoring. The presentation is available from his website at parse.com. He went into the technical details of how to accomplish this type of monitoring and what equipment you need. If you are handy with a soldering iron, you’ll like this one. Interesting and enjoyable - especially the anecdote about the guys who cracked the glass meter housing on his house and he had caught it on the monitoring camera. It does make for a very strong bargaining position when asking them to fix the problem.

Fourth Session - The silent network

Peter Hansteen gave a talk on denying malware through the use of free tools. It started with a history lesson and brought us up to “today” for some acceptably recent value of today (things may have changed in the past week). The talk centered around spam, viruses and worms and the eradication/denial or effectively neutering their impact. An important item to note is that malware once caused harm, now it tries to be invisible so that it can do it’s job without your notice.

In order to not kill the server with too many scanner processes, techniques like greylisting/greytrapping are used. In combination with pf and spamd, the number of successfull receipts of bad mail go down significantly. This kind of noise in the email system can be effectively fought at the perimeter. Slides of an older presentation are available at Peters website. Hopefully we’ll have the BSDCan presentation available on the BSDCan website soon.

Fifth Session - The Varnish HTTP accelerator

Squid is ineffective, something new is required. At least that’s how it starts. Modern web services are loosely based on CMS technologies, so a page is the result of a number of database queries and basically has dynamic content - so much for a caching engine like squid. Squid is an old design and does not take into account modern computers and the whole virtualized hardware/software paradigm that is in use today. Varnish is simply an accelerator. It uses RAM as opposed to disk (io is expensive) and has some CMS related features. It is designed to work with the current trend.

The talk went into detail about the places where we could gain efficiency and made comparisons to squid. The numbers are very impressive and the graphs show some pretty persuasive curves. As this is the meat of the talk, viewing the actual paper would  be of more value than my abbreviated gloss. You can get it from the varnish roadshow. It really is worth looking over if you have a site that sits behind a cach of soem kind (or even if it doesn’t). This isn’t a silver bullet, but it may save you some very serious cycles with a high volume site.

BSDcert Beta

I signed up for this just to see how much I needed to learn for the actual certification. I haven’t touched BSD in any serious fashion since the early eighties, so this was an eye-opener. There was 3 hours allocated, but most peoiple were out after about 1.5 hours. Given the list of people who participated, that’s not surprising. I am interested in seeing my results. I’m guessing at about 50%, which isn’t too bad given the design of the exam. Practical knowledge in other unices should be worth something.

It’s been a long day and I need sleep. It starts afresh tomorrow.

[Since I’m editing this a few days later (well, over a week later), It’s still been a long day and I am tired again, so Day four waits another day. I also notice that I haven’t mentioned the new toys.]