> > It is entirely possible to share those resources in a
> > read-only manner across SYSPLEX boundaries. It is equally
> > certain that writing to a dataset (improperly) shared in that
> > manner will result in a broken dataset and grief to the
> > consumers on both sides. Ask me how I know....

Dave said
>     You either really know by looking at the "code" or you broke it on
> some occasion or other :)  Or both <vbg> and maybe on purpose <mbG)

I have never seen the code and I have never been curious enough to take
a dump and figure it out in detail, but I have had conversations with
current and former (IBM) developers so I have a general idea of how the
technology works. Most of the senior ISV folks who mess around in this
area probably do too.

As for the scars... well in a former life, I and my team started using
program objects, which required using PDSE libraries, which required our
IT folks and our QA folks and our packaging folks to come to grips with
PDSE datasets for the first time.

It was a miserable experience. Most of the people involved refused to
believe they really couldn't share across SYSPLEX boundaries because it
often looked like it worked. If you IPLed the systems before the changed
datasets were used then sometimes it worked and sometimes it didn't. And
if you ever wrote to the dataset from two different systems you got a
broken dataset every single time. Guess who got the job of figuring it
all out?

Other folks crashed and burned because they failed to recognize when
things really were shared. The common operational practice of copying
new (executable) library members into a read-only dataset that was
shared by two or more SYSPLEX instances did not seem (to them) as a
violation of the sharing rules. And of course the number of dead or
dying systems did not seem to be a clue...

CC

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html

Reply via email to