Jan wrote:
In this kind of discussions, you have believers and non-believers (whatever
arguments are advanced by the other clan),
it's like a religious war.
I have 3 questions.
Q1. Are you a sysplex-er which went the ICF-only way (by applying (more than
probably) System- or User-Managed Duplexing?
If yes, can we get in touch with you to exchange experiences?
Q2. Is the table on p.27 of Dave's paper about approximate impact ratios (multipliers) relative to the simplex cost & the
percentage of > CF operations to be applied for each category OK? Or did you apply (slight) adjustements to those
multipliers?
Q3. Appendix A outlined a methodology to calculate the Simplex resource use. Is there a tool commonly available? In stead of each >
having to write it's own spreadsheet in it's little corner?
And when it would be available, I suppose it contains the duplex aspect with
the multipli
Jan,
I'm just jumping in because nobody answered (or I missed the reply):
Well, from my personal view it is sort of an academic discussion coming to
costs.
Guessing you are in a HA environment (nobody else would discuss this subject
seriosly ;-),
it really depends on your workload/config. I'm very sorry about this answer,
BUT it really depends.
- What would be your standalone CF-Config? (# of Processors, Links/Type,
Storage, CF-Level, Coexistence requirements a.s.o)
- Please post some metrics (RMF)
But to give you some general advice:
- My last project was a heavy SAP/R3 DB2(V7)-Server-Config (>40 server instances each in HA config) integration project running on
two z/990's with 8 Lpars's (1xProd, 1xQA, 1xSystec-Test, 1 ICF (1 ICP, 24GB) each), coupled by 4 (later 6) ISC-3 peermode links in a
single sysplex in CF/SD mode for more than 2 years.
In DB2 V7 there are some restrictions regarding to signalling in failover situations, even V8 does not lookalike to be finished, and
this does apply to CF's at all regardless of the Standalone/ICF question.
So, what is your mission critical application (Subsystem). It is really more a matter of sysplex exploiters and (eventually) then
FOLLOWING a discussion on Standalone/ICF philosophy.
Sorry (intenionally) not ansering Q2+Q3 but this is too early for the info's
you did provide.
HTH
Peter
----------------------------------------------------------------------
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
