Jump to content

Project Shining to Improve Avatar and Object Streaming Speeds


Vick Forcella
 Share

You are about to reply to a thread that has been inactive for 4310 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

"the regolith in the Linden Lab secret lunar base" Torley?

 

New serverfarm with HQ speedy thingies: Great!


Dropping a farm: Great, the closer they are together the better!


Sunshine: I have seen the behaviour when an avarar only rezzes *after* the server has a good chat with the computer of the avatar. By making this Serverside it will improve a lot, both the bystanders as the avatar will see instantly what the server knows. It will reduce the load on a connection. That is specially important when the user is on a slow connection. Next to that it should reduce the load on the LL servers that is the bottleneck when it comes to connections.

Perhaps this can be improved in such a way that it's possible to have SL run on a tablet with streaming fragments of the world.


Caching: Finally some good use for that small cache we have on our machines. SL doesn't speed up with having set a large cache, the opposite, it slows down. This action is very welcomed.


HTTP Library: Unsure. HTTP has lost us residents and the ones that figured out ho to set UDP again could get affected by this. I have no idea what HTTP LIbrary is supposed to be.

Link to comment
Share on other sites

All excellent news, especially object caching - and they should make a substantial difference to load times and the like.

With such good news around (and many other improvements that LL haven't mentioned) I do wonder why we don't get a headlline, less techie, article.  Come on LL - talk up Second life a bit more.

Link to comment
Share on other sites

This all sounds good but the fact remains, things have not improved or at best have improved very little, or have improved and then been destabilized again by more changes.  This couple with pasted LL attempts at grand fixes followed by huge disasters leaves me anything but optimistic, but we shall see.  It is a pity though this information was not released via this forum where more would see it, for at least it does give us some idea of what LL is attempting to do at the moment and that even if not reassuring is good to know.  Finally, let me state the obvious in the end whether rezzing or other operations are server or viewer based the key information comes from the core servers and if they are nto working, and it is very obvious these are not working well and have not for a long time now we will have trouble.  For those of you that can't understand what i am saying consider this example, I log out in zerango and log in at my home in Turnip, do I expect zerango to send my information to turnip or is that stored in the core server which then sends it out to whatever mechanism is in place to handle it in Turnip.  Right now I see a lot of smoke, a few mirrors, and very little that I can call positive action, though if the present situation continues for a week or so and is not destroyed by this weeks restart I might be ready to conside that some minor improvments have occurred.  Now though I am not optimistic as i have been seeing these promises for six years now and have seen very few postive results come from them.

 

Link to comment
Share on other sites

Avatar: Hurray! Now people on bad connections won't show up as clouds for extended amounts of time =)

Object Cache: Hopefully the cache will be bigger, too. I suspect the use of e-tags just as browsers use to cache data. I'm really looking forward to this one. It should make playing over wireless better, too.

HTTP: Will you be implementing SPDY? It seems a good time to do that as well, and it will speed things even more!

This all sounds smart from a corporate view, too. Bandwidth is a variable, expensive cost. By making these changes the Lab should reduce its operating expenses. 

Link to comment
Share on other sites


Annalena Yoshikawa wrote:

It is Googles  http 2.0

Thanks for the link. As I know Linden Lab I place my bet on them making their own implementation. That way they can tweak without having to worry about copy-write stuff, and making connections more secure as a side effect.

Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 4310 days.

Please take a moment to consider if this thread is worth bumping.

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share

×
×
  • Create New...