Peering Inside: The Second Life year in review (part 2)
Stability and bugs
While Linden Lab's quality of service metrics indicate that everything's been on the improvement more or less, most Second Life users will tell you that particularly the last six months have been a whole lot of trouble. Previously the grid tended to have two basic operational modes: Working and not working. You might have server, network or client lag on top of that.
Nowadays we have partial subsystem failures. That is, the grid and simulators tend to remain running even when supporting systems are having a cow, responding badly, or failing to respond at all. We're not really quite sure if that constitutes an improvement or not, on the whole, considering how poor the ongoing experience can be, with daily subsystem hassles, erratic behavior, and the inevitable subtle viewer bugs. Rather than just having the show go belly-up when something fluttered, the show now goes on, exposing your face to the whirling jagged edges of razor-sharp bugs and explosive database failures.
If we recounted all the downtime incidents, faults, problems and bugs for this year in one place, our editors would probably freak out. Their number is beyond sensible and reasonable reckoning.
Clearly a lot of work has been done on reliability and stability, and just as clearly, there is still a long way to go.