On Thu, 3 Mar 2005 10:04:09 -0000, "SPP Admin"
<mre@yahoodonteventhinkaboutspam.com> wrote:

Apart from the obvious problems that Berkley are having with their servers 
and power, isn't there another problem that Seti has, given that it is 
clearly the most popular Distributed Computing project.

The hardware they have can clearly only just about cope with the number of 
users - thus when something goes wrong - i.e. a power surge takes things 
down temporarily, they are completely overwhelmed when they come back online 
with the number of people trying to send data in - which in turn, either 
fouls up the server again, or delays users sending their data in.

In this context, and for the good of Distributed Computing and Seti in 
particular and science in general, is it not better that the number of seti 
users is reduced - allowing more data to be returned more successfully and 
the Berkely to catch up with the backlog, completing the move to Boinc.

It's not that we would be asking people to move off DC projects completely - 
there are plenty of other projects that could use the computer power, just 
that we give Seti a breathing space.

In this context, and because of the frustration of having a stalled 
screensaver, after over 4 years with Seti, and 1970 processed unites, I've 
moved to Einstein@Home and the Boinc project. When Seti's better, I'll 
probably make it one of my other boinc projects, but I think it needs some 
space right now!

Uh, that is NOT the problem! The Boinc\Seti section of the project has
about 75,000 users TOTAL! And a bunch of those have never even sent in
a single unit! The new machine can handle a LOT more since it is only
running at less than 10% capacity when the project is fully up and
running. It can handle 2 more cpus and 4 more gig of memory. The
problem is that everytime Berkeley tries to do something, a peripheal
event occurs that throws a huge monkey wrench into the process. They
got brand new machines and they were faulty! They got brand new memory
and it was bad. They got brand new hard drives and they caused
problems. They used new programming code, as suggested by the machine
and OS suppliers, and things didn't go as they were told they would.
They tried doing a backup as the manufacturer of the hardware told
them too and the machine crashed. Turns out the manufacturer forgot
something. They are also trying to run the project on a TERRABYTE
database! That means that the data, when it is manupilated, is
EXTENSIVE! Not a simple 2 minute job.
Then for some, as yet, unknown reason the power switches off. Then it
happens AGAIN!!! Now the manufacturer of the machines is sending them
a UPS program that will gracefully shutdown the machines when the UPS
comes on, but "forgot" to tell them it needs "special" cables!
It is a case of trying to keep EVERYONE happy while at the same time
trying to keep within a VERY small budget! They do an EXCELLENT job!
Leaving the project is NOT the best way BUT crunching for other
projects while Berkeley is having problems IS a good idea!
Remember they are in the process of shutting down Classic too! When
that happens, who knows how soon, there WILL be extra hardware
available for Boinc\Seti use. AND less pressure because then they will
have only one project to maintain instead of the current two.