[sldev] Exciting reuter's post about OpenSim - project to create open source version of SL Servers

[Laurent Laborde]

On 9/8/07, Tim Shephard <tshephard at gmail.com> wrote:
> http://secondlife.reuters.com/stories/2007/09/06/rival-grids-threaten-lindens-monopoly-on-sl-technology/
>
> For those who are looking for something to test their Clients against,
> this might be the way to go.

I don't like how this article is written, biased.
The main problem about secondlife is scalability.

In fact, SecondLife official servers and OpenSim have exactly the same problems.

Linden Lab have some troubles with the huge central asset servers.
I copy/paste opensim : "While there's no centralized inventory server,
meaning that an avatar on DeepGrid can't take objects from one region
into another, users can cross region boundaries seamlessly,
experiencing no disruption as their client connects to servers on
opposite sides of the world."

Understand : "We don't have any problem with this features, because
it's not implemented".

If i remember correctly, the sim crossing problem appeared when the
grid started to grow very quickly.

"About 300 servers have installed Frisby's open source Second Life
server code, called OpenSim." ""Second Life is a game of Russian
Roulette with the login server." ...
My fact : "DeepGrid is a game of Russian Roulette to find a running server".

I'm not against OpenSim nor LindenLab servers, i like both, they are
just not playing on the same playground... they are just ... not ...
competitors. E.g : OpenSim could be very nice for intranet or small
scale grid.

If you read the reuter's article without I.T knowledge you just read :
"OpenSim did better in 2 months than Linden Lab in many year". And i
just can't accept that, it's just wrong.

I worked as syadmin on heavy load clusters. Not the kind of HPC
supercluster with very well defined workload, but the kind of cluster
to handle angry mob with never-ending feature adding and fixing.

And i don't blame LL for their problem, as i don't blame coders when
they say "I assumed that the world's fastest hardware will be fast
enough to handle this very simple data logging. So, yes, our 10
millions users cluster is down because of a race condition in this
standard logging library"... See what i mean ?

-- 
kerunix Flan