As long as we are throwing opinions in here, I'll add mine. And you get what you have paid for it. :-)
Depending on the actual scenario, I have seen cases where both Mark and Patrick are correct (measurement). But if the truth be known (belief), most managers would manage to the $$$ bottom line, so the overhead of CP switching is an acceptable compromise regardless of its impact (belief, but backed by actual experience). John's point about software costs echo mine. zAAPs were initially acquired because shops didn't want CPU-hogging JAVA work to monopolize CPs under any circumstances. It wasn't about JAVA performing well; it was about them not affecting other work. As JAVA on z/OS has become more pervasive, its performance became important and the paradigm has shifted slightly. But costs are still the yardstick most shops configure to. Reviewing zIIP and zAAP eligible time relative to the S4HRA of MSUs is normally a sufficient cost justification for acquiring additional specialty engines *because* while we entertain thoughts of cheap GP CPs, we ain't there yet! :-) Bob -----Original Message----- From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf Of McKown, John Sent: Thursday, February 04, 2010 11:05 AM To: [email protected] Subject: Re: IBM countersues Neon over zPrime accelerator We have a z9BC-T02 (soon to be downgraded to a Q02). A zAAP will definately out perform a CP in our circumstance. Well, for Java work, that is. If we had any. So, it is true that a unstricted CP and a zAAP run at the same speed. Therefore, in the situation where software costs are not relevant, an unretricted CP is better, more peformant, than a zAAP. Even for zAAP eligable work! Now, how many people here think that software costs are not relevant? Don't be shy! So, even if you have unrestricted CPs, a zAAP may be more "cost effective" than a CP. For zAAP eligible work, that is. zAAPs cost less to acquire than CPs. zAAPS don't add to your software bill. So, if while it may well be true that a CP is "better" when looked at in a "pure performance" scenario, for TCO purposes a zAAP may be a far better choice. -- John McKown Systems Engineer IV IT ---------------------------------------------------------------------- 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
