Journal Entries
Merry Christmas
Posted Dec 25, 2003
Merry Christmas to all. I'm sure that, like me, all your thoughts are about that visitor from the heavens arriving to bring good news to all men.
Unless, of course, Beagle 2 crashes and burns...
Best of luck, guys. I would be up watching it live but the BBC doesn't appear to scheduling a live 'Sky at Night' in honour of this event, and I expect the kids will be up early in the morning, so I'll keep my fingers crossed anyway.
Discuss this Journal entry [5]
Latest reply: Dec 25, 2003
I think we're making progress
Posted Oct 23, 2003
As one or two of you may have noticed, h2g2 (and DNA in general) has been a tad, well, slow in recent days. And a bit flaky.
Most of this is down to the speed of the database. Adding new features inevitably means making more requests to the database, and some requests run slower than others. I've spent most of this week staring at performance graphs and page after page of profiling information, to try to spot those areas where we're performing badly. But we seem to have got somewhere, finally. Our most recent patch went up just before 8pm tonight, and almost immediately the CPU usage dropped from close to 100% most of the time to more like 25%.
One of the biggest culprits was Search. That's always been a query that takes a long time, but looking at the profile output, it soon became clear that things would be going quite quickly, then there would be a patch of very long running queries, and in among those queries (some taking minutes to run) was usually one or two searches. So it seemed that Search was blocking other queries. Changing the query to not lock other tables should help with that.
It's often difficult to tell if these changes make a difference. Sometimes you'll apply a patch, things calm down a bit, but then just get busy again later. But this time, something has definitely made a difference. Let's see if it lasts.
Discuss this Journal entry [39]
Latest reply: Oct 23, 2003
Doctor Who is coming back
Posted Sep 26, 2003
http://news.bbc.co.uk/1/hi/entertainment/3140786.stm
It's odd. This news should have filled me with an almost transcendant joy but... Perhaps I've been disappointed too many times and I'm now convinced that it's never going to happen.
However, it really does now look like it will happen. Jimster (who's known about this for ages, the scamp) was positively bouncing when I saw him earlier. And the fact that it's Russell T Davies writing it makes me very optimistic - I've been wanting him to write Doctor Who ever since I saw 'Dark Season' so I'm confident he'll do a fine job.
Who knows - if they spend the same kind of money on it that they spend on (say) Walking with Dinosaurs then it might even look good.
I guess the campaign to cast your favourite actor as the Doctor starts here.
It's doubly strange that this news should break today, but unfortunately, I'm sworn to secrecy about the other news I heard this morning, which is not entirely unrelated to Doctor Who...
Discuss this Journal entry [6]
Latest reply: Sep 26, 2003
I'm so very, very sorry
Posted Sep 10, 2003
To all those inconvenienced by the DNA servers taking unofficial strike action in the last hour, I have to apologise profusely.
I had asked our server people to run some SQL monitoring on the database server, in order to get a live snapshot of what queries were happening when, and which ones were running slowly (so that I can try and improve those areas which are tending to run slowly based on real data, and not just guesses). Unfortunately, it would appear that this monitoring was taking up so much server resource itself that it caused the database server to grind to a halt.
Once this was corrected, and all the web servers were restarted, everything sprang back to normal.
So it was all my fault, and I promise not to do it again.
And it looks like I'm back to guessing...
Discuss this Journal entry [16]
Latest reply: Sep 10, 2003
Announcement: Server upgrades and load balancing
Posted Sep 5, 2003
I just thought I'd announce that we're now running on a set of new servers. Four new web servers have been added to the pool of DNA servers, which means we now have eight in total, although currently we're only running on the four new ones. The old ones will be re-added to the pool shortly.
We've also changed the way the servers are load-balanced. Until now, our servers were balanced in quite a simple fashion. The BBC has a large pool of web servers - these are what serve up most of the pages on www.bbc.co.uk. When you first ask for a page from bbc.co.uk, the response will come from one of these servers, selected mostly randomly, but also based on where you are (we have servers in New York as well as London). But for applications like DNA, which need a bit more processing power than just flat HTML, we pass the request on to a dedicated server.
Until now, each front-end BBC server has been linked directly to one of the DNA servers. This means that if one of the BBC servers gets a large spike in traffic, that spike is passed directly to the DNA server behind it, meaning that the server will struggle and occasionally fail.
Now, however, we have switched to selecting DNA servers at random. This means that each time you request a page, you might get the response from a different server to the one you talked to last time.
The reason for doing this is to try and smooth out load spikes. In situations where one server would get hammered, now that load will be spread over four (and shortly, eight) servers, which means that, in theory at least, the service should be more predictable and reliable.
As I say, that's the theory. But in any event, double the servers and smoothed loading can only help.
Discuss this Journal entry [47]
Latest reply: Sep 5, 2003
Jim Lynn
Researcher U6
Write an Entry
"The Hitchhiker's Guide to the Galaxy is a wholly remarkable book. It has been compiled and recompiled many times and under many different editorships. It contains contributions from countless numbers of travellers and researchers."