Last week was not a good week. This coming week is shaping up to be pretty hectic (though I am hoping not as personally unpleasant), so there may be a blogging slowdown.
I spent all day yesterday at the 2013 edition of the Texas LINUX Fest. I haven’t been since 2010, but that had less to do with my frustrations with the 2010 organization and more to do with personal issues. (In 2011, that just turned out to be a bad weekend, with having to get my car inspected and deal with other things. Last year, it was in San Antonio; while that may be a welcome change of pace, the schedule wasn’t compelling enough to make me drive 150 miles round trip.)
I thought about doing detailed summaries of each session I attended, but frankly I’m a little worn out and a little lazy. I’d rather mention a handful of panels I did like. (There were some others that I went to, but don’t feel I can fairly evaluate because they weren’t what I was expecting, or I was distracted by other issues (see below), or, in one case, I just think it’d be a jerk move to badmouth the presenter.)
I really enjoyed Theo Schlossnagle’s “Scaling: Lessons Learned and Their Applications to Apache Culture” keynote speech, which covered a lot of good points about complex systems. He sees commonalities between building scalable systems and building communities to support them. Many of the points he made may not be hot news flashes but are worth repeating. Among those:
- People get so caught up in how awesome it it to build stuff that they forget what the real world looks like.
- Code is just a tool. It isn’t a child or a family member. You don’t have loyalty to it.
- Engineers have a tendency to focus on the technology they love instead of the actual problems they face.
- At the core of things, your job is to tell the computer what to do.
- Unbalanced hyperspecialization leads to poorly constructed solutions.
- The biggest challenge is that increasing scale and increasing performance demands lead to increased complexity.
- Technological complexity is an emergent property of complex and changing business problems. This complexity has to be understood and managed, which is difficult for specialists.
- If you don’t provide value, your [stuff] doesn’t matter.
- In order to survive, we need generalists. Schlossnagle didn’t quote Heinlein, but he might as well have.
David Stokes from Oracle did what I thought was an excellent talk on “The Proper Care and Feeding of a MySQL Database”. I’ve mentioned before that I’ve been dabbling in MySQL, so I got a lot out of this. Some of it may have been obvious (more RAM, more disks, good things. Use decent hardware, not something you scavenged from the admin assistant because it was too slow to run the latest Office), but the two things we learn from history are that too many people don’t learn from history, and that the obvious often isn’t.
Philip Ballew’s “Ubuntu; Where We Were, and Where We Are” presentation was…amusing, shall we say, mostly for the level of skepticism directed at Ballew from the audience, many of whom seem to be skeptical about recent Ubuntu decisions like the replacement of X. I’m becoming increasingly skeptical of Ubuntu myself; I just upgraded to 13.04, and now I’m running into the “The system is running in low-graphics mode” error, which I haven’t had time to fully debug. The worst part is that I’m getting this only intermittently; I think it may be a timing issue, possibly with some Virtual Box kernel extensions.
Owen Delong’s “IPv6 – It’s Easy on LINUX” presentation was also very good. I haven’t even started to configure my systems for IPv6 (and I’m not sure everything supports it: I’m sure about the Mac and Project e, but less sure about some older gear), so I found Delong’s talk useful. I was surprised, though, that there was even more hostility and skepticism from the crowd than there was at the Ubuntu panel. Why is IPv6 an issue in 2013? And many of the questions from the crowd seemed to boil down to “How do I emulate this particular thing I do in IPv4 using IPv6, even though the reason this is needed in IPv4 is because we have a limited number of IPv4 addresses available, where in IPv6 we could give every single atom in the universe a unique address and not run out?”
Okay, that was a long question, but you get the point.
Brad Richardson’s “GPU based password recovery on LINUX” lightning talk is worth checking out. He was able to do the talk in about five minutes, instead of the allotted ten, and the subject is interesting; using reasonably priced GPUs, you can rapidly break MD5 hashes, orders of magnitude faster than throwing a general purpose CPU at the problem. (Richardson’s slides give specific performance figures: try 16 hours 46 minutes to brute-force a “8 character password with lowercase, uppercase, and numbers”, versus an estimated 36 days for a CPU based attack.)
Anyway. Tomorrow is the start (for me) of Yet Another Perl Conference 2013. (I registered for the conference itself, but couldn’t afford any of the training going on over the weekend or after the conference. Plus the training conflicted with the LINUX Fest.) I expect to be pretty tied up Monday through Wednesday, though I will try to blog from YAPC as downtime and network connectivity permits. I may even try to blog YAPC 2013 itself, but I can’t promise that.
Edited to add: Why did I not have a “Perl” category on this blog, but did have a “Python” category, given that I use Perl more often than Python? Fixed.
Edited to add 2: Thinking some more about it, it made sense to have a “Programming Languages” category and make Perl, Python, and others sub-categories below that. I’m still thinking about whether it makes sense to put the languages category under “CompSci”, but that way lies
Edited to add 3: I realized there were two other points I wanted to make.
- I was much more favorably impressed with the organization of TXLF this year than I was in 2010. Of course, they’ve had four of these, so you would expect them to have the bugs fixed. Still, I was impressed at how smoothly almost everything from registration onwards ran. The only problem I saw was an unexplained 20 minute delay in the start of the lightning talks, but I didn’t feel that was a major issue.
- The quantity of tchotchkes available at TXLF? Very high. The quality of tchotchkes available? Still evaluating that, but I’m decently impressed. Favorites: the microfiber cleaning cloths from OrangeFS, and the SavvisDirect USB/12V adapters. Special mention goes to Hostgator, who were giving away a much wider variety of tchotchkes than any other single vendor.