RSS Feed for This PostCurrent Article

Second Life: Failure Waiting to Happen




SL FailureIt is Monday morning here in New Zealand, Sunday in the U.S., and I just tried to log in to Second Life.  But Second Life had 67,333 current logins and could handle no more load.  They were at their user limit!  C’mon, how un-scalable is this platform?

I recently asked Linden Labs how I could get a full refund for the University sim that we purchased, and I cited the highly unreliable performance of the system as the reason for my request.  Over the course of several weeks, I noted dozens of instances where the system failed, almost daily.  Now, Linden Labs may cite that they were up and running, but if teleporting doesn’t work, or clothes don’t download, or object libraries are unavailable, or certain regions are undergoing un-announced ‘maintenance’ and can’t be reached, then for a class meeting the service is a complete system failure. If we (educators) are to utilise environments like Second Life for teaching, we must have assurances that service will be available on a reliable basis.  Reliable doesn’t mean 80% available, it means 99.99%.  Industry standard.  This is crucial for Linden to understand.  It is nearly impossible to reschedule a cancelled class meeting during busy semesters, so if a class is cancelled we simply lose that time.  It is unrecoverable teaching time.   

In my request to Linden Labs I asked that we first have a discussion about a Service Level Agreement (SLA) to see if there might be a way forward.  To my shock I receive a reply from one of the do-not-reply-to ‘Lindens’ (can’t they do business in the real world?) asking me to clarify what I meant by ‘Service Level’.  Have these folks never heard of ITIL?  Again, remarkable.  Well, for those Lindens who are still sitting there with a “huh?” on their face, here is a link that will explain ITIL “…delivery and control activities to achieve operational excellence on a day-to-day basis” – Service Operation.  Read it.

I hate being so seemingly impatient, but most educators are passionate about the precious time allocated to interact with students – we cannot tolerate shoddy support systems.  These frequent SL failures cause me great concern for the future use of such systems for real education.  The great promise of high-fidelity graphics, voice and animation means nothing if reliability isn’t present.  Importantly, reliability is a prerequisite.  It must be present in the first instance, not developed after Linden experiment to see if there is community interest in the product .  Do what is necessary to provide up-front Service Level reliability, or fail.  Full stop.  Given the fact that there are dozens of such environments either in production or in development, I fear that Linden’s offering is failure waiting to happen…it may be just a matter of time.

Trackback URL

Create a free edublog to get your own comment avatar (and more!)

  1. 4 Comment(s)

  2.   By brent on Oct 22, 2008 | Reply

    Hmmm… i hate to say I told you so, but “I told you so.” Any centralized “platform” is always going to suffer from these kind of scalability issues if it really takes off. Where Linden Labs made a huge mistake in my opinion was in gratuitously pandering to the Open Source bandwagon by Open Sourcing the client! What they should have done was Open Sourced the server. That way we could have had Virtual Worlds distributed across servers all over the place … umm… much like … oh yea! the ineternet.

  3.   By Scott Diener on Oct 22, 2008 | Reply

    Yes, you have often “told me so”. But objectively, not all centralised platforms suffer from this (see Croquet Consortium). Our OpenSim collaboration with Otago, Canterbury and Telecom is probably more to the issue that you are addressing, however. This National Virtual World Grid project utilises the OpenSim platform and the Second Life viewer. And I’ve already tested “teleporting” between these platforms. Slow progress, but promising.

    Nonetheless, I admit that I agree with the underlying premiss you suggest here. Dang, I hate having to do that. :-) . Warmest regards.

  4.   By brent on Oct 25, 2008 | Reply

    I suppose not; google for example could be argued to be essentially a ‘centralized platform’ for many things at this point and it does quite well — it probably makes a bit more $ than Linden Labs does though. I thought that the whole point of Open Croquet was that essentially distributed in a kind of peer-to-peer model — which is more scalable as more people are added to the grid? I always thought that Linden Labs should have Open Sourced the server — that way we could have had SL servers running all over the place to run the thing and Linded Labs would have still had the first mover advantage and been able to capitalize on this .. instead they’ve kept it mostly behind closed doors and i’d be surprised if they make it in the long run.

  5.   By Scott Diener on Oct 26, 2008 | Reply

    Exactly, as is Amazon with their AWS capabilities for scaling. The OpenSim sever looks/acts identical to Second Life, and is likely to establish the distributed server space to which you refer. IBM is championing this…take note of their amazing OpenSim development – http://www.beyondspaceandtime.org.

    Time will tell about the different marketing strategies (closed or open source)….

Post a Comment

*
To prove you're a person (not a spam script), type the security word shown in the picture.
Anti-Spam Image