We have implemented a Bronzeplex that sounds like the same thing you are
thinking of.  We wrote an ISGNQXIT routine that has worked without trouble
for a couple of years now.  There are several considerations in ISGNQXIT
that I can help you with, if you're interested we can do that in email.
If you are on or considering 1.6, we found a couple of GRS bugs, one is
a new APAR, I can look up the number.  And the other is an 0C4 in an FRR,
they haven't gotten far enough to assign an APAR number yet.
A couple of our people transferred to the IBM testing organization.  They
looked at what we had done with ISGNQXIT, and what was found with the APAR,
and they decided that a Bronzeplex configuration was too far outside a normal configuration to include in their testing. Which was interesting,
but if you're considering Bronzeplex, you ought to be aware.

Lance


----- Original Message ----- From: <[EMAIL PROTECTED]>
Newsgroups: bit.listserv.ibm-main
Sent: Thursday, August 25, 2005 3:25 PM
Subject: Merging SYSPLEXes


We are looking at merging 2 SYSPLEXes which are 20 KMs apart. Our first
step is to create a Bronze-PLEX as described in SG26-6818 Merging
Systems into a SYSPLEX. The driver is to achieve single SYSPLEX
pricing. We want to create on SYSPLEX where all systems would share the
same ISGLOCK structure and SYSPLEX CDS datasets but the 2 original
SYSPLEXes would continue to use their own set of DASD volumes. The
systems of each original SYSPLEX would continue to use separate control
datasets for products like HSM, ACF2, CA1, CA-Solve etc.
We understand that there will be issues related to duplicate dataset
names and volsers and that there will be some performance impact. What
we are most interested in is advise on potential enqueue deadlocks that
may be introduced for products which share a control dataset. We would
have 2 copies of the control dataset (one for each of the original
SYSPLEXes) but the enqueues would be seen by all systems in the new
merged SYSPLEX.
The products that this applies to are DFHSM, SMS, ACF2, CA1, Control-M,
Control-D, CA-Solve, HSC.
We are also looking at ISGNQXIT as it can modify resource names which
may be useful in specific cases. Anybody used that? Another idea is to
capture the enqueue workload from SYSPLEX A and redrive it on SYSPLEX
B. Anyone done this?
Any comments, recommendations or details of problems experienced would
be greatly appreciated.


----------------------------------------------------------------------
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