Stan writes:
>Yep.  The bigger the file, the worse the performance hit.
 
>>From my viewpoint:
 
>  1) this is a bug in FUNLOCK ... it should *not* be posting pages to
>     disk;
 
>  2) it was probably put into FUNLOCK as a workaround to some other
>     problem in MPE (my assumption).
 
I don't think this is a bug. The reason is that FUNLOCK has always had
the effect of flushing buffers to disc; it makes sense when buffers
are private to a process as under MPE/V. For many programs, therefore,
FUNLOCK has had the additional effect of forcing all I/O out after
a transaction has completed. While the buffer synchronization problem
is no longer a concern, eliminating the forcing of data
out to disc would have the side effect of making some formerly robust
programs much more sensitive to system crashes.
 
Still, it would be nice to have the option.
 
-- Bruce
[log in to unmask]
OPT, Inc.