Hi John,

We (I) had an idea similar to this several years back and now we are already 
sort of doing this with both our SyzCMD/z and SyzMPF/z products.  They allow 
the user (or via automation) to create a memory area that we assign the 
name/token pair to, then we allow that area to be used by any other part of our 
software.  (At first) It was great, and it gave us 16 whole bytes :)  and it 
was user-select-able to be persistent for just the "job" that created it (or on 
who's behalf it was created) or until the next IPL.  As I mentioned, it used to 
be limited to that 16 byte area, but when several customers said they wanted "a 
little bit" more, we made the size (almost) unlimited by playing with the way 
IBM handled the name/tokens.  But as that space used is still within ECSA, you 
still have to be 'reasonable' on size, so we placed a arbitrary user- 
select-able limit of anywhere between 128 bytes to 2K limit on the spaces we 
created as well as the "total" space that can be used in this way (I think we 
limit it to no more than 20 MB total or to within 20% of exhausting their 
available ECSA (80% used max) (whichever is lower)).  We now have a new version 
that we are getting ready to ship (in February or March) that has added the 
capability of keeping this area in a "Space" (sort of like shared temp data 
space/hiper space) and it allows us to take away most of the limits on size.  
It requires a special little started task that owns the space and does all of 
the real work, and incidentally makes management relatively simple.  It's not 
quite as fast as having the data in ECSA (and we still allow, and actually 
prefer clients use that option), but has the advantage of being allowed to be 
much bigger.   

I think that the "hard" part was working out how to do the clean-up side of 
things and also how to make sure that we 1) didn't let someone get carried away 
and fill up ECSA and 2) have a way to, both on demand and periodically, perform 
a "sanity check" or "garbage collection" so that we don't leave stuff sitting 
around that should no longer be there.  That necessitated adding an additional 
option when the data was created (besides JOB duration and permanent duration) 
that was a "Variable" duration which the creator can use to set an "expiration 
limit" such that even if the JOB had not yet completed (some jobs can run for 
months), or in the case where the data needed to be semi-permanent, that we had 
a way to remove it "semi-automatically".  

We are also providing an "sort of" API that will allow the sites to access this 
data from within their own code.  I want to call it an API because that makes 
it sound sexy, but truly it's just a called program.

So, while it doesn't do exactly what you were looking for, it does allow that 
type of functionality.  At some point we are hoping to be able to allow direct 
access to the storage space, but we are probably still a ways off on that, and 
I have no idea who would even have a need for it right now.  It would just be 
kind of cool.

Brian Westerman
Syzygy Incorporated

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to