Wow, Sun is acquiring MySQL. That’s sort of stunning.

My first impression is that MySQL has been under a bit of pressure in that big companies like IBM and Oracle have been buying up the companies who provide their transactional storage engines. Now MySQL has the (perhaps diminishing) clout of Sun behind them. Oh, and suddenly Sun goes from relatively irrelevant in the Web development market to incredibly relevant.

And yes, I know that Sun provides Java, but the JDK is only one small piece of the overall Java Web application platform. The language is the language, but what affects developers most is the IDE, deployment platform, application server, and the libraries and frameworks that they use, and many Java developers don’t depend on Sun for any of that stuff.

I am a heavy user of MySQL, so I hope that this turns out well.

Update: Blog reactions from all over. Check out Simon Phipps, Jonathan Schwartz, and Jeremy Cole.

In World of Warcraft, there are raid instances that reset once a week. The instances consist of several encounters that usually have to be defeated in order. For example, Gruul’s Lair, the easiest 25 man raid instance, has encounters with High King Maulgar and then Gruul. The hardest instance, the Battle for Mount Hyjal, includes five encounters.

Playing all of these encounters takes time. The fights themselves are time consuming, and Blizzard includes “trash” monsters between each of them to increase the overall time it takes to complete the instance. In each instance, the encounters get progressively more difficult. So teams that can kill the one boss cannot necessarily defeat the one that follows. In game terms, this is what progression is all about. Every time you win an encounter, you get more stuff to enable you to move on to the next boss.

What winds up happening, though, is that teams (usually guilds) progress as far as they can and the instances reset at the end of the week, leaving some hard bosses alive. This creates a market opportunity for those guilds in that they can sell their unfinished instances to guilds further along in their progression. The sellers can’t kill those bosses anyway, and the buyers get the opportunity to spend their time on their own progression rather than killing bosses that don’t drop anything they really need.

I find it really interesting to see economics provide a solution to a common in-game problem, which is that doing the same raids over and over is time consuming and to be honest, incredibly boring. Progression-oriented guilds need the loot that the later bosses in raid instances that they’ve already conquered drop to complete high end armor sets and upgrade their weapons, but to continue their progression they need to avoid spending entire evenings rehashing content they’ve mastered. The market delivers.

There are a lot of old Web sites out there these days. Plenty of organizations have had a presence on the Web for upwards of ten years, and like it or not, their Web sites are showing their age. It seems like I run into more and more groups who have old, crufty Web sites running many applications written in a variety of languages. What I’m wondering is what the best option is for these organizations is in terms of Web hosting?

If they are just running static HTML and maybe some PHP or CGI applications, then shared hosting is perfect. They just need a place to upload files and perhaps Web-based database administration. Shared hosting is cheap, and it’s up to the hosting provider to make sure the servers aren’t running a some Russian mafia member’s IRC server to control a zombie army.

Organizations with enough invested in the Web to have their own systems administrators are in OK shape, as well. They can go with managed hosting, or colocation, or even just buy some connectivity and run the servers out of their office. Their budget is large enough to pay somebody to make sure their Web sites work, so they’re really beyond the scope of what I’m talking about.

I’m talking about those people in the middle. Maybe they have a relatively low traffic Web site that has some old CGI scripts, a Java Web application, and a ColdFusion application or two. Their setup is too complicated for shared hosting, but they don’t have the staff or the budget to take on managing their own Web servers. What’s the best option for them right now? Do they need to subscribe to managed Web hosting and then outsource their system administration to an outfit like Pythian?

It seems like what I see most often for these groups is that they have their own server running out of date hardware and infrequently patched software, being tended by the person in the office who’s in charge of making sure that people can get email and that their antivirus software is up to date. Usually these servers are down a lot and are sitting ducks for whatever malware happens to cross their paths. Surely there must be a better way.

Feedback much appreciated.

For all of the negative talk about Rails lately, the truth is that for cranking out Web applications quickly and easily, it remains difficult to beat. I started on a new project last week, and in an afternoon, I managed to go from having nothing but a CSV export of a database to having a fully implemented data model of 6 or 7 tables, the model classes to represent them, unit tests for those models, and a script that imports the data from the CSV file to populate the database.

Today I managed to get full text search of the site working (with all of the dependent objects indexed) in about six hours, using Sphinx, Ultrasphinx, and will_paginate, despite the fact that I had never used Sphinx or Ultrasphinx before, and that I had to patch Ultrasphinx to make it work the way I wanted. I think that next time I could probably do the Sphinx implementation in an hour or two, even for a complex application.

Yeah, Rails is hard to set up in shared hosting, and maybe some people are sick of Rails being the flavor of the month, but the bottom line is that even though I feel a bit rusty with Rails, I was still able to be significantly more productive than I would have been with PHP or Java.

Tonight I happened to catch the end of the Golden Globes press event, on which two of the stupidest people in the world read the names of the award winners. The real show was canceled because the actors agreed not to cross the writers’ picket line. Anyway, in addition to reading the lists of nominees and then announcing the winners, the presenters of the show also treated us to some analysis and commentary along the way. A low point was reached when immediately before the host opened the envelope to read the winner of the “best picture” award, he asked the resident expert to predict the winner. The pointlessness of that prediction made me think of the pointlessness of nearly all predictions.

Forecasting is important. If you’re thinking of buying a house, it’s good to have an idea of what the housing market is going to do in your area and whether mortgage rates are going to rise or fall in the immediate future. If you’re trying to choose which company to work for, it’s important to try to figure out what the fortunes are of the companies you’re considering. Sports fans are interested in whether their favorite team is going to get better or worse next season. I’m all for forecasting. What I’m sick of is predictions.

For the past couple of weeks we’ve seen every political pundit in the world make an ass of themselves predicting what would happen in the Iowa and New Hampshire primaries. Now that the NFL playoffs are under way, we are forced to watch football pundits predict what’s going to happen in a game we’re about to watch, and then to listen to them make more predictions about the rest of the game during halftime. In the technology industry, we’re being bombarded with predictions of what Steve Jobs is going to announce in his Macworld keynote on Tuesday.

The sad thing is that nearly all of these predictions are utter and complete self indulgence. People make predictions so that they can pat themselves on the back if they’re right, and pretend like they were right if they were wrong. If predictions do have any utility, it’s in the evidence cited to substantiate a prediction. The prediction still won’t be enlightening, but at least the discussion of the prediction will be. All too often, that evidence is left undiscussed by the experts or “experts” making the predictions.

Please, let’s have more forecasting, but I’ve had it with the predictions. That said, I forecast that this blog post will not have a measurable effect on the number of predictions that people make.

I’m one of the people who was concerned that the portrayal of the characters in the Baltimore Sun newsroom in the season premiere of The Wire was a tad unsubtle. Actual newspaper employee Fred Clark disagrees:

I’ve read reviews complaining that Haynes seems unrealistically heroic, or that the newsroom scenes are overly explicit, with the lines of right and wrong drawn in uncharacteristically broad strokes. We’ll see. That can happen in a series’ final season, when the writers sometimes feel a now-or-never pressure that can override a previously subtle touch. But to me the scenes in the newsroom were painfully real. The dialogue rang true with words and phrases I’ve heard — or said myself — dozens of times.

The whole post is worth reading, as it explains the David Simon ethos of craftsmen versus careerists. Great stuff. If you, like me, don’t know what a “local information center” is, you definitely need to read it.