OS/400 version running BPCS version -- BPCS-L

We will soon be upgrading our AS/436 box, a IBM 9402-436 RISC-2102, which runs
BPCS 405 CD, from V3R7M0 to V4R3 & since there seems to be a lot of interest
in performance issues on BPCS_L - I thought it might be smart to mention this.

I lack the performance expertise that I am seeing on BPCS_L and also I am
still a newcomer to OS/400.  We are relying heavily on our local hardware
support KSAI to see us through the transition.

Here is some info to put our site in perspective.

Our BPCS Y2K consultants did add a bit of SQL to the BPCS 405 CD mix.

We are running some S/36 code re-worked for OS/400.

Machine/36 is alive with intermittent visitors.

Total on-line users is just under 30 with several sessions each.

We abandoned e-commerce (EDI I) to get the Y2K conversion accomplished.

We are about 2/3 Green Screen & 1/3 Emulation.

I have been working at Central Industries since August 1984 - it is the second
longest I have been with any one employer & as I am now aged 55 I would just
as soon stay here for the rest of my useful career.

We have 3 physical sites connected via SDLC - one is pure IBM hardware, the
other has Perle cluster of remote controllers attached to a Motorola Modem -
the Motorola is under IBM maintenance & we also have a spare one in a closet,
after an experience with lightning frying an IBM modem & replacement not being
super fast.  That was not IBM fault - one of our sites is on a sandy soil
flood plain that does not have an electrical ground & at the time of modem
going down it was surrounded by a violent union strike that lasted 3 years.

When we were running the business on SSP, I did run SMF occasionally & it
pretty much confirmed my overall perceptions as to where our bottlenecks were,
but there were some issues it was helpful clarifying.  e.g. Would a faster
comm line help a particular site - answer only 10% of the time does something
have to wait due to comm bottleneck, but as we grow sessions & printers there,
they will need faster comm & in the meantime, we have a disk I/O problem, in
which I have balanced stuff across drives as best I can, but more memory for
cache will make a big difference.

We got more memory for cache, using EMC brand name, & it was a big help - IBM
had said that if the hit-miss ratio on cache was 2-1 or better, then cache was
working effectively to our benefit & we generally got 7-1, but on my home PC I
often get 20-1 - I had tried various cache flavors & I do have an ancient box
that needs serious upgrade.

We got Acceler8 from ASNA & it was a big help with BPCS/36 & we did serious
benchmarks before deciding to keep it, because we had also tried it years ago
on MAPICS I & it did more harm there than good.

We also got a faster comm line for the growing site & it was a big help, to
which I partially attribute growing the comm line speed faster than the demand
on bandwidth.

Faster processors & CISC-RISC & better OS/4 version also was big help.

The biggest leap was about 50 fold increase in speed when we went from SSP on
5360 to SSP on OS/400, in 2 installments - AS/436 then guest OS.

There are times that I think that the relatively low price investment in
something better from IBM is a bigger bang for the buck for the company, than
me futzing around with performance analysis.

Al Macintyre
+---
| This is the BPCS Users Mailing List!
| To submit a new message, send your mail to BPCS-L@midrange.com.
| To subscribe to this list send email to BPCS-L-SUB@midrange.com.
| To unsubscribe from this list send email to BPCS-L-UNSUB@midrange.com.
| Questions should be directed to the list owner: dasmussen@aol.com
+---