Introducing z/OS Capacity Provisioning

Performance and capacity management on z Systems must ensure that work is processed according to the service level agreements that are in place. Guaranteeing service levels continues to be a relatively static task as long as the workloads being considered are sufficiently stable. However, in many environments, workloads fluctuate considerably over time. Because the total workload or mixture of workloads varies, it can become increasingly difficult to guarantee service levels. With z/OS® Workload Management (WLM), incoming work is classified with a performance goal and a business priority. WLM tries to accommodate the goals of all the work in the system. However, even with an ideal WLM service definition, it might not be possible to achieve all specified goals when the total workload increases. In this case, trade-offs must be made. WLM decides which goals can be compromised first, based on the assigned importance level. Discretionary work is displaced first, followed by low importance work.

At some point, however, this solution might not be acceptable, either because the displaced work is relevant from a business perspective, or because it interacts with resources that are required by more important work. There can be different reasons for this situation to occur with different ways how they can be resolved.

  1. The CPC's available processor capacity has been exhausted.

    If the processor capacity of your zEnterprise server is insufficient the capacity may need to be increased to accommodate the increased workload. This can require a permanent capacity increase for planned growth, or a temporary capacity increase for seasonal or unpredictable peak periods. IBM® z Systems can quickly and non-disruptively activate additional processor capacity. That capability is built directly into z Systems servers. It is provided by IBM Capacity Upgrade on Demand (CUoD) for a permanent increase in processor capability, and IBM On/Off Capacity on Demand (On/Off CoD) for a temporary capacity increase.

    On/Off Capacity on Demand allows the configuration, for example, of general purpose processor (CP) capacity and specialty processors, such as zAAPs, zIIPs, IFLs, ICFs, or SAPs. Several models of z Systems servers are subcapacity models. On such models, additional general purpose processor capacity can be provided by a different capacity level, additional processors, or a combination of the two.

    For example, for the z10 EC server, the capacity levels for the CP engine are 7, 6, 5, and 4. Full capacity CP engine is indicated by 7; subcapacity CP engines are indicated by 6, 5, and 4. For the z10 BC server, the capacity levels for the CP engine are A-Z. A full capacity CP engine is indicated by the letter Z. The capacity setting is derived from both the capacity level and the number of CPs. For the z10 EC server, the capacity settings are 4xx, 5xx, 6xx, 7xx. For the z10 BC server, the capacity levels are Axx - Zxx.

  2. When Defined Capacity or Group Capacity has been exceeded.

    Such a situation can occur because the available processor capacity has been exhausted or due to WLM soft capping when Workload License Charges (WLC) are in effect and the Defined Capacity or Group Capacity limit has been exceeded. In this situation, your zEnterprise server might still have enough processor capacity but WLM caps the capacity available to one or more images for cost reasons. To overcome this situation you can temporarily increase Defined Capacity or Group Capacity.

  3. Logical processor resources are insufficient.

    This situation may arise while your zEnterprise Server still has sufficient processor capacity but the number of logical processors of a partition is insufficient to contain the workload. The available physical capacity cannot be used by the partition that needs it. Since there is still sufficient processor capacity on the server, this capacity can be directed to the workload by configuring additional logical processors online.

For more information about CPC and LPAR configuration and capacity planning, you can search for these topics on IBM Techdocs.

In the context of capacity provisioning, the capacity of a z/OS system means processor capacity. A z/OS system can consume different types of processor capacity:
  • General purpose capacity

    Processor capacity of general purpose processors (CPs), usually measured in million service units (MSU) per hour, or just MSU

  • Application assist processor capacity

    Processor capacity of System z Application Assist Processors (zAAP), measured in number of processors

  • Integrated information processor capacity

    Processor capacity of System z Integrated Information Processors (zIIP), measured in number of processors