I've noticed recently that there's a common theme that occurs in life, or at least, my life. I'll do something really expensive, e.g. move house, go on holiday, walk past a chocolate shop etc. and then be completely surprised when something necessarily expensive happens soon afterwards. This appears to be happening at the moment, where most of the expense of moving house, sorting out bills etc is hitting my just at the time when I want to start thinking about saving for Christmas. As a result, I've worked out how much I can allow myself to spend in November and it's got a minus sign in front of it. Not good.
Still, this time of year does have various pleasantries that make it worthwhile. I went to Alexandra Palace last night to see a rather good fireworks thingie. Being billed as "the biggest display London's ever seen", it was a little amusing that during the display, you could see the slightly bigger and more expensive display going on at Victoria (I think) in the background. At least I know where to go next year.
I've also been having email issues. A while back, I decided that running my own email server-thing was far too complicated and that I should just leave it down to Google. No more having to back up email, no more playing with spam filters, GMail would sort everything out for me. This was especially attractive given the rumour that Google were soon to provide IMAP capabilities to GMail. I was happy, for about a month. Then the spam started to roll in. I eventually arrived at the conclusion that GMail's spam filtering is crap. I get a *lot* of spam per day (about 3,500 messages in a last few days alone) and evidently GMail isn't good enough to be able to filter out really obvious stuff. So I packed it in, and gone back to running my own server. I might have to spend more frustrating time fixing it when it breaks, but at least I can tune the spam filter once and then forget about it.
On another note, I'm doing some research into Lighttpd. I'm using that at home on virtual boxes with practically no memory and it's hugely faster than apache. Might try and host this site on there and run some tests.
Not a lot has been happening recently. And I really can't be bothered to write anything exciting. That said, I seem to be doing a lot of waiting at the moment.
The good news is that I'm planning the Christmas theme properly this year, so hopefully it'll be GOOD! Although I'll probably decide it's awful on December 2nd and replace with something else. We'll see.
It may amuse some to learn that I've started taking driving lessons. Hurrah! Having spent the last three mornings dribbling around roads near Stockport I can safely say that this driving thing isn't half as difficult as I thought it would be. I've a mild tendency to nearly crash into things, but I'm sure that will happily evolve into a habit of actually crashing into things, given enough time. All I can say is that it's now a long time since I was 17 and it's about time.
In other news an interesting thing happened at work. Whilst building an ASP.NET application designed to serve the needs of a few thousand people, we (the team! go team!) were considering ways in which the implementation costs could be cut down. Being a big(ish) company, we have things called "standards" which the cynics amongst us would claim only exist to get in everyone's way. Of course, it's actually a good idea to only have a few selected bits of technology and platforms on which everything is built, because then it's a lot easier to support, but I digress.
The "standard" in this case was to build the database on Microsoft SQL Server. Now, unfortunately, this is about $20k per CPU. Given our habit of building things a) twice and b) on big hardware, this would have been an outlay of (2x 4-way boxes) $160k. This is a lot of money, easily dwarfing the amount that needed to be spent on trivial things like, hardware, people, that kind of thing. So we needed an alternative. As far as I knew, the only other standard was Oracle, which is ridiculous. So, throwing caution to the wind, I put the whole thing on MySQL running on RedHat and didn't tell anyone.
I was surprised. It ran a lot quicker. I didn't quite believe this, so I did some tests using JMeter. Surprisingly, I found that under load, the MySQL backend gave about a 3-fold performance improvement to the application over the MSSQL backend. I should point out that not much else changed - the db schema was the same, with the same triggers, stored procedures and indexes etc. Now, some people may say that this says a lot about my ability to tune a MySQL database versus a MSSQL database and they're probably right. However, I'm not a database expert and managed to get the MySQL db running a lot faster using only the internet (specifically the fabulous site: The MySQL performance blog) and a bit of guesswork. To me, that says a lot.
As it happens, the hand of the standards police has pointed at me, and it turns out that PostgreSQL on Solaris is also a standard. This has the advantage of being both cheap and supported, so I now have to once again go through a lengthy migration process to move everything over. Hopefully I'll learn something in the meantime.
So, the moral of the story is: Don't listen to people. Mostly.