> > > > However, with PLPA overflowing into Common the majority of PLPA page-ins > > will come from the Common Page Dataset, not PLPA. You will have > contention > > for the two exposures when concurrent page-in requests for PLPA happen > at > > the same time as Common Page-in/out activity using the same two > exposures. > > I don't think so, there are 2 exposures per pagedataset, not per paging > type. So if PLPA pages must come from the Common pgds, they will use the 2 > Common pgds exposures and must compete with the normal Common pageins from > the Common pgds.
Vernoy, read it again. I said: "the majority of PLPA page-ins will come from the Common Page Dataset" I may not have said it clearly, but we are saying the same thing. It cannot satisfy a PLPA page in from the Common page dataset if it is handling Common pages from the same dataset. > > > At the same time the PLPA dataset will be relatively inactive. > > And will also be the 2 PLPA pgds exposures. My Point exactly!!! > > maximum potential page-in rate is achieved by using the maximum number of > exposures and this must mean to route PLPA pagins through the PLPA > exposures > to the PLPA pgds and the same for the Common. > However, since most installations do (should) not have (any) PLPA paging, > the benefit might be small and possibly not worth while. > Believe it or not Vernoy, you are supporting my point - With WLM PAV put Common and PLPA on the same volume with WLM managed PAV; and without PAV Common and PLPA on different volumes. In either case there should be no overflow for maximum exposures. I also agree that most systems do not have PLPA or Common paging, but when it does happen it really, really hurts. We all no the impact a small DPR from Aux has on the CICS TCB, so just think about what happens when many addresses are waiting on the same page faults to be resolved from PLPA or Common. A Local page fault usually affects one ASID, whereas Common and PLPA page faults usually affect many ASID. To quote Ted - "been there; done that; got the T-shirt; don't wanna go back." Ron ---------------------------------------------------------------------- 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
