You're right Ted. In fact, I really dont want to share CPs with the CF. However, the current soft-papping mechanism only allows to soft-cap an LPAR to a certain amount of MSUs. At first I used soft-cap. Then I noticed we encounter situations where one LPAR is being soft-capped, while all the others are not even close to their defined capacity. From my point of view, this is a poor allocation of resources (which we are paying for). Obviously, If the LPAR was not soft-capped (and not hard-capped) it could have used the free MSUs.
I would like to see some solution to allow soft-capping the entire box at a certain amount of MSUs, and then define regular weights to the LPARs. That way, if an LPAR is not hard-capped it can use free MSUs (if there are any), and still not go over the defined capacity limit for the box. Maybe something similar to the way they limit CP processing power on the z/890, only controlled dynamically. Gil. On 12/31/05, Ted MacNEIL <[EMAIL PROTECTED]> wrote: > > You don't really want to share CF CP's, still. > Also, MVS Busy is a rough estimate of latent demand. > Every time an image is pre-empted, LPAR assumes that you wanted the rest > of (or > the next) time-slice. So, it is added to the MVS Busy field. > If you are not pre-empted, then nothing is added. > > Any kind of capping is going to introduce a performance penalty. > With soft-capping, you have to decide if the penalty is worth the software > savings. > If you are meeting SLA's (and saving money), don't worry, be happy! > -teD > Me? A skeptic? I trust you have proof! > > ---------------------------------------------------------------------- > 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 > ---------------------------------------------------------------------- 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
