LHC@home

Basicaly, with the 1/10 size work units, the webservers are handling 10x the number of client requests/stat tracking traffic.


______________________________________________________________
Did your tech wear a static strap? No? Well, there ya go! :p

> Basicaly, with the 1/10 size work units, the webservers are handling 10x the
> number of client requests/stat tracking traffic.

Is that why within the last day, it's only taken me 40 minutes as opposed to 6 hours and 45 minutes to go through a WU?

> Is that why within the last day, it's only taken me 40 minutes as opposed to 6
> hours and 45 minutes to go through a WU?
>

Probably ;-) Previous units simulated 1M turns, while current do 100k.
I am still afraid, that with estimated number of 50000 users, LHC may expect similar problems to S@H

These shorter work units have nothing to gain from more runs than they already got. Althought they are good as a load test, we try to get rid of them and hope that we never see them again:) (This was server administrator's point of view. Our physicists may hold different point of view.)



Markku Degerholm
LHC@home Admin

We will know if they are going to have the same problems as SETI@Home when they goto the same number of users and get the same load on the servers (assuming the same/similar hardware).

Though, the back end load still may not be the same due to the different validators and so forth ...

That is the one thing about server loads, they rarely scale on a linear basis after some particular "knee" in the load ... Like you usually can never see 100 Mbit on a 100 Mbit network ...
<p>
For BOINC Documentaion: Click Me!