Yesterday, Meta Linden posted the Service Quality Metrics of November 2007 in a post in the Official Linden Blog. Meta did an analysis of the data (not really self-critic), but i’d like to do mine. Of course, i’m not an expert in networked systems, just a customer of Linden Lab, but even being just this i think i can have my opinion (or maybe because i’m this, someone who pays Linden Lab).

Let’s see first some of this SQM graphs by Linden Lab:

Second Life Service Outage Graph

Second Life Percentage of Viewer Sessions Sessions Graph

First, i’d like to comment the Outage Graph, the first one.

Since the start of the Het-Grid project the planned outage is getting down (good thing) but the unplanned one is getting really high, specially in July, August and November. With unplanned time loss percentage arriving sometimes to almost 2% i think this Het-Grid is not working as good as it was supposed to work. It seems more than Linden Lab is replacing Planned for Unplanned outage. And this is BAD!

About the second graph, the Viewer Sessions Crashed one. As you can see the percentage of Viewer Session Crashed is… OMFG!!! from 21 to 24%!!!! so almost 1/4 of the started sessions crash!!!! (well, not far from reality, 1/4 of my sessions crash normally)

Is Second Life a Beta??? Well with this level of crashes it seems a Beta. I presume this graph talks about the “normal” viewer, not the First Look Windlight or the Havoc Beta one (is quite normal this other viewers crash, they are clearly beta releases).

Linden Lab is getting better on some things like communication with the residents (specially good communication about the “Quick Time Exploid” problem), but they need to work in stability and performance more, much more.

Share this:
  • Twitter
  • del.icio.us
  • Technorati
  • Google Bookmarks
  • Ping.fm
  • Digg
  • Facebook
  • StumbleUpon
  • Tumblr