Filed under: News and developments, Trends by Digado

This weekend Orange Island organized a meeting in Second Live of a range of topics, and called it the ‘Orange Geekend’. I was present at an interesting (technical) update by PhD John Plevyak on scalability of Virtual Worlds in the future. Thanks to Simbian Chronos for re-reading my report of this because some of the really technical details went right over my head.

Scaling Virtual Worlds is more complicated than simply increasing server capacity. When envisioning a world with thousands of avatars at one place at a time there is more to be done. John explained there are two ways to be able to get about 10.000-50.000 avatars in one place at the same time and still run smoothly. The first is Prioritizing. Check the client for prioritized targets to receive updates. With 50.000 avatars constantly sending out information you need to know what messages matter. The second solution is making some data Client rendered, such as in most MMORPG’s, reducing server strain by skipping various verification processes such as location, movement, physics, animation etc. but keeping the truly important parts such as money, transactions and communication serverside.

The second part of Johns talk was about Trends, the most important one being peer 2 peer sharing of CPU. Shared CPU will be able to support the Java and .Net security backends by moving part of LSL processes into the client. Another way of moving the data bottleneck from the server to the clients was the use of a ‘Ghost Client’ called ‘Afterlife’ currently in early alpha in the labs of Metaversatility. These ghost clients communicate with the maingrid, but cause no extra server stress due to the scene of a concert for instance, being ‘broadcasted’ by a number of real avatars on the spot to these Ghost Clients.

In short: The focus of hardware development for the coming years for Virtual Worlds will be on shifting the load from the server side, to the client. Which means as much as your computer will have to do more of the heavy lifting instead of the servers over at Linden Labs. This makes the system more scalable, as the more avatars there are, the more computers are logged in to share the heavy lifting of processing scripts and transferring data.

Obviously this is crucial from a marketing point of view. We often talk about a ‘tipping point’ in a number of users where a second Hype Cycle of Second Life could start, say for instance 20 million sign ups or 2 million active users. If the hardware and server capacity can’t support these numbers, there is no hope of growth, as people will leave due to lag or constantly full sims at populair locations, creating a circle of attracting and repelling people without significant growth.

Posted on: January 26, 2008 | Share on: