Wednesday, August 13, 2008


Recently, Linden Lab hired Frank Ambrose, who apparently is something of an expert with managing and growing networks. The general consensus and statements from the Labs themselves indicate that this is all to increase scalability.

On an interesting tangent, M Linden note a slowdown of growth and new users which has also seemed to factor into this.

All I have to say is that Linden Lab is finally noticing scalability is a problem? The upper boundary of SL concurrency was about 45,000 when I started in November '06. These days, about two years later, it's 63,000 or thereabouts. That seems like respectable growth, except that during the time concurrency climbed up about twenty thousand users, the overall number of users increased by the millions.

Now, concurrency is a fickle beast to figure. Seeing the figure 63,000 may seem low, but that number is just the number online in that point in time. 63,000 at 9 pm may be 49,000 at 10 pm, or it may be a completely different set of 63,000 people. So it is not easy to accurately determine just how much damage (if any) concurrency is causing. We could assume that every hour there is always 60,000 users online, and every hour it is a completely different set of 60,000 people. 60,000 at 24 hours in a day, means that in the course of a day that gross and very shabby estimate is 1,440,000 people get into Second Life.

The next step in our horrible guessitmations is that the total amount of user accounts (be it bot, one day wonder, or hardened oldbie) is way past that number. I know this because I do recall noting around December 2006 that there were more people registered than the population of most American cities (my own included, which was why it was memorable). Even assuming one quarter of those registrations are accounts that log in more than one day (bots and oldies), that's still too many people to log in all in one day. The count according to the blog is 14,000,000.

Why do I include bots? Simple, because they log in with the rest of us. A bot is, as far as Second Life is concerned (speculation on my part) the same as a person. Both log in and suck resources from the SL servers (when the Lindens aren't spilling coffee on them).

Perhaps you are saying to yourself, "but surely those million or so regular users aren't logging in all at once or so, the number of residents who logged in the last 60 or so days is only a million, less than that in the past week!" Consider that statistic, though. We established that we grossly rounded the concurrency and determined that only 1.5 million people can get online on a day with top concurrency. In the real world, it drops massively from time to time and so never reaches this theoretical (and shifty) value.

The point, yes let's get to the point, I am trying to make is that concurrency isn't supporting Second Life. There have been increasing problems with inventory and regions failing and such, and Linden Lab has on occasion shut SL for new log-ins (the recent outcry was that they were targeting freebie accounts, denying them the 'right' to log in). Concurrency is failing. For some time now, the major limit has been that SL starts to get excited when it approaches upwards of 55,000 online at a time. And while this upper limit is growing due to the ceaseless efforts of the Lindens, it's not supporting the mass of users behind it.

It took them long enough to figure this out.

Now, some people are shouting, "Rejoice! For they may finally improve stability!" Well, that depends. When they say scalability, do they mean 'the number of people we can cram into here' or 'improving server performance so a cleaner and better experience is had at higher concurrency, thus raising the bar'?

The former sounds just like Linden Lab, doesn't it? I will give them some credit. M Linden has talked more talk than Philip ever did in his last few years as CEO of LL. And his speeches have certainly bent around a different idea, the user experience instead of 'how cool is this???!'.

But for some reason, I just can't see that happening. I think everyone is getting their hopes up. My gut is telling me that they are just looking to cram more into less. I want to believe. I WANT TO BELIEVE.

Maybe they'll prove me dead wrong. I hope so.

No comments: