Second Life Looks At Scaling Problems

News.com reports that Second Life is looking down the barrel of some major scaling problems as the virtual world's popularity soars. While Linden Labs itself seems confident in the scalabilty of their virtuality, outsiders aren't so sure. From the article: "'My understanding of (Linden Lab's) back-end requirements are that they're absurd and unsustainable,' said Daniel James, CEO of Three Rings, publisher of the online game Puzzle Pirates. 'They have (about) as many peak simultaneous players as we do, and we're doing it on four CPUs.'"

Is this a point of failure?

[grub]

FTA: Each server is responsible for an individual "sim," or 16 acres of virtual "Second Life" land. At peak usage that means that each server is handling about three users."

That sounds like a point of failure. What would happen if a large group of mischievous users organize and decide to visit the same 16 acres of land simultaneously?

Re:Is this a point of failure?

[ThosLives]

If you read the article, it's clear that each server isn't limited to three users: it means that the system is such that the number of peak users divided by the number of servers is 3. I'm sure that they have more than three users in a given 16-acre area at any given time. (16 acres is fairly large, actually).

The big difference with SL

[joshv]

The big difference with Second Life (SL), which I am sure is hugely 'wasteful' of server CPU, is scripting. Any user can create an object and attach one or more scripts to it. These scripts can communicate with other objects, with Avatars, change the state of the object, create new objects, listen for input, etc... It is not unusually to see a sophisticated Avatar with many hundred object attachments constituting its 'custom'. Each of these objects can have an embedded script, which runs on the server. That's one Avatar. Now think about ten such Avatars congregating in once place, all of these objects interactive with each other, and the each of the Avatars in real time. This requires a lot of processing power, and loads of inter-object communication.

LSL (the name of the scripting language) is interpretted, and oriented towards simplicity, not speed. It's goal is to make it relatively easy for users to create richly interactive and complex objects, at the cost of server CPU.

So yes, per Avatar, I am sure SL uses more server CPU than any other massively multiplayer game, but that's because it offers an extremely rich and customizable experience.

It scales *financially*

[Anonymous Coward]

The things that distinguish this setup are that

a) They charge users per acre of virtual world that they 'own'. The article claims that the monthly income from 16 acres is $200 - which is gonna pay for a cheap bare-bones server in just a few months. As their user base goes up - so does their income - and so they buy more servers.

b) Remember that user scripts continue to run even when there is nobody inside the 16 acre patch to see it happen. There was a /. story a few days ago about someone who built a system of plants, pollinating insects, clouds and rain in their little patch of the virtual world. Even when that person isn't logged in - and nobody is watching, those acres of virtual world still have to run the weather/insect/plant modelling. It's possible that in this system, the bulk of the compute power is NOT in running users - but in running the world. If that's the case then they may not care about how many people crowd into one area because it's only the owner of that area that's putting out a bazillion scripts.

I think this is a sustainable business model - so long as the amount and complexity of scripting that can be crammed into each 16 acre plot stays within reasonable limits - and I think they can handle that.

Another concern would be if people started leaving the community in droves - then there might be areas of land (ie servers) with only one person homesteading there. This kind of fragmentation of user location might result in the need to run thousands of servers that - whilst paid for as capital assets - are no longer generating enough revenue to pay for their ongoing upkeep. This could probably be handled by forcably relocating people to consolidate acreage - but that might not be too popular.

But all the while the community is growing and thriving - I doubt they have a problem.

Moores law

[captainclever]

Moores law will certainly feature in their business plan :)