Re: Queries

["David M. Malon" <malon@anl.gov>]

>  > >  I don't know about the EXECTIME... David Z. ?
>  > >  I was going to use 30 seconds for -t (processing time)...
>  > >  what do I need to use?
>  > >  thanks.
>  > >  --Alex
>  > >
>  > 
>  >    What you use for -t depends on whether you're running the tests Arie
>  >    called TEST A, where caching time is expected to dominate.  Since some of
>  >    the files are <10 MB, 30 seconds per event will dominate caching time.
>  >    I suggest -t 0 or -t .01 for the A tests.
>  > 
>  >    David
>  
>  I see....
>  then, after you examine the qeuries that I just sent to gcdev, will you
>  recommend
>  processing time for each?
>  Thanks...
>  --Alex

   Here is what I would do:  Run the tests as though you were doing Test A 
   (cache time dominates)--use -t 0.01 or something like that.  Then 
   examine the file delivery times logged by the CM, and you'll have a good 
   idea what values of -t will 

      a) cause processing time to dominate, or (for smaller t)
      b) lead to a mix of caching-vs-processing domination.

   David