Recent Comments

    IBM Soft-Capping is a feature which allows controlling the WLC bill by setting a limit for each LPAR of a MAXIMUM for the average consumption of MSU (R4H). This limit is called Defined Capacity (DC).

    The Soft-Capping rule is when the Rolling 4 Hours (R4H) becomes superior or equal to DC then the LPAR is capped. That means that the IMSU consumption will not be able to exceed the DC anymore until the R4H becomes lower than the DC.

    Figure 1: the LPAR01 billing level cannot reach 110 MSU but the IMSU is brought back to 110 MSU (DC) when the soft-capping starts, so when the R4H reach 110 MSU.

    Because DC is a fixed limit, we face a lack of flexibility. During capping period the LPAR is constrained while other LPAR have a highest DC than necessary for their consumption and these reserved MSU cannot be used by the capped LPAR.

    GCL

    Introduced by IBM from zOS.1.8 and z9 + µcode, Group Capacity Limit (GCL) allows to set a capacity limit for a group of LPARs. When the sum of the LPARs’ R4Hs in the group reaches the GCL, all LPARs in the group are capped at the level of their share% weight in the group (and not at the levle of their DC). It allows to simplify the soft-capping management but the good thing is GCL is basic and the bad thing is GCL is basic too because of not using white space and because of the use of the weight for the capping by PR/SM. This metric is used for other things as Hard-Capping or guaranty of MSU when the server is 100%busy, so the value of the weight can be not at the right value at the right time during group capping. Here we face the same lack of flexibility as standard soft-capping.

    Restrictions (*): Defined and Group Capacity

    • A partition must not be defined with dedicated processors and a partition must be defined with shared processors and WAIT Completion = NO

    Hard Capping (Initial Capping) must not be defined

    • z/OS must not run as a VM guest

    • PR/SM capping works within ±3.6% from the defined capping value

     

    * IBM Germany Research & Development , 2010, August