Hi Fotis, * --
I enjoyed your posting thoroughly! You make an excellent candidate for
the Stephen Burke award ;-)
A couple of comments:
Fotis Georgatos wrote:
> During the submissions, I managed to confirm that the RBs have a breaking
> point at around 32-64 concurrent submissions of jobs and a throughput of 25
> jobs/minute. The exact threshold of submission failures might be
Worth remembering that these figures depend on many many things. For
example, if you increase the number of sites, or make the JDL more
complicated, your numbers will change. Try including a required match
of a bunch of logical files ...
> These sites are excessively loaded, if compared to the few cpus they
> provide;
> This is typically a result of massive submissions that are done with the -r
> parameter, which is only a problem if people don't realise they can take
> advantage of the grid :-)
I know what you mean ... but on the other hand, massive submissions done
WITHOUT the -r parameter can also generate skewed job distributions. It
is possible that what you observe is due to normal submission via the
RB coupled with bad site information -- the incompleteness of the glue
schema and the resulting limitations of the info providers are well
known. OR what you observe could be due (as you said) to people using -r
because they think they can do better than the RB+info providers.
J "hope this post passes the Markus Constructive Email test" T
|