My story is improbable but true. We have four full function parallel
sysplexes in our enterprise. Of these four, three were converted to star
mode some time ago. The fourth had at most two members, one of which was
very lightly used. So lightly used, in fact, that we stopped running it
altogether; the LPAR has not even been activated for some time. Conversion
from ring to star mode hardly seemed worth the trouble.

A few months ago we had occasion to run a catalog scan job on all
sysplexes. It did very heavy reads but no updates. We noticed after a while
that the job ran *many times longer* on the ring mode sysplex than anywhere
else. No more catalogs to scan; no more data sets; just a lot more elapsed
time. Only one image IPLed on a huge box: a 2064-1C6 with 10 gig of memory.

I had an inspiration. We converted the sysplex to star mode. From that
moment on, with no other changes, the catalog job ran as well there as it
did on the other star mode sysplexes.

Conclusion: even with *one* member, star mode rules.

.
.
.
JO.Skip Robinson
Southern California Edison Company
SHARE MVS Program Co-Manager
626-302-7535 Office
323-715-0595 Mobile
[EMAIL PROTECTED]

IBM Mainframe Discussion List <[email protected]> wrote on 11/04/2005
03:31:48 AM:

> I created sandbox sysplex within one CPC (shared CPs for both MVS and CF
> LPARs, IC channels).
> I observed significant performance improvement when changed GRS from
> ring to star mode. I didn't notice such improvement, when changed XCF
> transport from CTC (ESCON) to CF structures.
>  From the other hand I was told that there is not noticeable difference
> between ring mode and star mode for two-member sysplex.
>
> Is the difference because of star mode advantage or just I have
> something something to tune when using ring mode ?
> BTW: I don't want to use ring, it's just curiosity.

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