Re: Parallel processing design for service

#1

Look at the n-step algorithm used in CPU task scheduling. It is the
what CPU's use today to schedule the execution of work. You could get
an idea from its design.

--
Bryan Phillips
MCSD, MCDBA, MCSE
Blog: http://bphillips76.spaces.live.com

"Ben" <Ben (AT) discussions (DOT) microsoft.com> wrote

Quote:

I'm currently writing a service that we will want to throttle for some
customers, so their database doesn't get hit too hard, but it's possible some
might have faster servers so will be able to get away with having more than
one process run at once, speeding response times.
Consequently I was planning to have the ability to set the maximum number of
processes that can be run at once, but enable this to be less than 1 - so
that if it's say 0.1 then if a process runs for 1 second then it has to wait
for 10 seconds before another can run.
The question I can't really decide though is should I do it like this with
just one flag, or should I have two flags - one for idle ratio and one for
number of parallel processes? If two, then should two processes at once add
on double the time for 'running' that one would, meaning it would have to
wait longer?

Has anybody else got any other novel ideas that could be implemented in this
situation?

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Re: Parallel processing design for service

Dotnet Distributed Applications microsoft.public.dotnet.distributed_apps

Re: Parallel processing design for service - 03-03-2007 , 08:05 AM