Engineering efforts

For me, one of the most interesting aspects of virtual worlds is the technology side. I'm a systems administrator myself, responsible for maintaining and optimizing an important internet application, and as such, I can appreciate the engineering efforts of my colleagues at Linden Labs, CCP et cetera. It's not an easy task to keep all those residents and pilots lag free and happy..

Linden Labs has long struggled with Second Life's performance. Over the years, there have been many improvements: as I am writing this, there are over 77.000 SL residents online, which is roughly about three times the number of concurrent residents usually seen in late 2006. And in those days, SL would crash if the number of active residents got near 30.000!

But today, it's CCP that I want to talk about. Eve Online (as Second Life, btw) is a single shard world, meaning all pilots inhabit the same instance of the virtual universe that is Eve Online. And some of solar systems put quite a strain on the servers! Take, for instance, Jita, the major trade hub of Eve Online, which is sometimes populated by 1000 pilots or more. Until recently, having that many pilots in local (in any one system) would mean intolerable lag, horrible performance and all the fun associated with it. Recently, however, CCP released a patch that drastically improved on that situation: bugs were fixed, code was refactored and a database index was added. It is now estimated that Jita can serve up to 1400 or perhaps even 1500 pilots. A considerable improvement!

CCP has published quite a detailed account of this effort on it's dev blog here. As an engineer, I really appreciate the level of detail they're willing to provide. If you're really into engineering efforts, you should read this one: debugging jita live is for real men, where CCP Atlas details how a critical bug was fixed on the Jita server - while the system was online. Quite an effort, and a good read for anyone involved in high availability internet application!