ACM concepts
ThruPut Manager AE monitors several different metrics, called “Limit Types”, compares them to limits set by the installation and defers subsets of the batch workload when appropriate. This results in reduced charges for software in both sub-capacity pricing and TFP environments, as well as lowered hardware costs when TFP-HW is in use.
ACM Limit Types
ACM defines four distinct Limit Types, each with their own configurable limits that an installation can use to determine whether a given limit type should cause batch workload constraints. The installation can choose to use any combination of one or more Limit Types or can use none.
R4HA
This limit type is used with sub-capacity pricing. ACM monitors the rolling 4-hour average (R4HA) for individual LPARs, LPAR Groups and ACM LPAR Sets which are described later in this manual. Country Multiplex Pricing (CMP) is also supported in which case ACM keeps track of the R4HA across multiple CPCs.
Purchased Capacity Percentage
This limit type is intended to be used in conjunction with IBM’s Tailored Fit Pricing Hardware Consumption Solution (TFP-HW). ACM maintains a rolling average of CPC MSU usage (default is 15 minutes). Every 10 seconds the average is updated and used to calculate a percentage of the current Purchased Capacity of the CPC. Should the Purchased Capacity be exceeded in a 15-minute interval, then the installation will be accumulating billable usage of the TFP-HW capacity corridor. TFP-HW uses fixed, not rolling 15-minute intervals, aligned on each quarter hour. ACM uses rolling intervals to provide better detection of usage trends.
The installation can alter the rolling average interval value specifically for this Limit Type to a selection of values ranging from 10 seconds to 15 minutes.
ACM calculates the Purchased Capacity of each CPC, defined as the portion of total capacity purchased by the installation that is currently assigned to the CPC. The Flexible Capacity for Cyber Resiliency feature of the IBM z16 allows capacity to be moved between z16 CPCs.
When configured, the TFP-HW corridor is the capacity immediately in excess of the Permanent Capacity of the CPC plus any active Flexible Capacity as applicable.
While primarily intended for controlling the use of the TFP-HW corridor, this Limit Type can also be used to control batch when any type of OOCoD (On/Off Capacity on Demand) is active.
CPC Busy Percentage
ACM monitors overall CPC general CP usage and maintains a rolling average of CPC Busy percentage (default is 1 minute), updating it every 10 seconds. High CPC usage increases measured MSU usage for identical workloads. This limit type can be used to constrain batch workload when CPC usage is high and provide savings in both sub-capacity and TFP environments.
The installation can alter the rolling average interval value specifically for CPC Busy Percentage tracking to a selection of values ranging from 10 seconds to 15 minutes. The shorter the interval, the more sensitive ACM will be to changes in CPC busy.
In the SLM policy, you can specify during installation if the CPC average is to be of only CP usage or combined CP and zIIP usage. An installation may want to combine CP and zIIP usage in a CPC with busy zIIP processors, particularly if the batch workload uses significant zIIP cycles.
MVS Busy Percentage
ACM monitors MVS Busy for the current LPAR and maintains a rolling average MVS Busy percentage (default is 1 minute), updating it every 10 seconds. During normal processing, only CP time is used. When a System Recovery Boost is active with a zIIP Boost, both CP and zIIP time is included. High MVS Busy values indicate LPAR overload and can increase measured MSU usage for identical workloads. This limit type can be used to constrain batch workload when the LPAR is overloaded and provide savings in both sub-capacity and TFP environments.
The installation can alter the rolling average interval value specifically for MVS Busy Percentage tracking to a selection of values ranging from 10 seconds to 15 minutes. The shorter the interval, the more sensitive ACM will be to changes in MVS Busy.
ACM Capacity Level
ACM does more than take actions only when a limit for a Limit Type is reached. It can act well before the limit is reached. You can configure ACM to gradually increase constraints on selected batch workloads as the limit is approached, and gradually decrease them as the usage decreases. This is accomplished by using a concept called the Capacity Level.
Capacity Level is a number that indicates how close to the limit the usage of a Limit Type is at present. The values range from 1 (highest usage) to 5. If the usage is below the value assigned to Capacity Level 5, then there is no active Capacity Level, and therefore no constraints on batch workload. You specify the mapping of usage level thresholds to Capacity Levels for each Limit Type in the SLM Policy.
ACM regularly calculates the Capacity Level for each Limit Type that has been enabled in the active SLM Policy. ACM picks the one that indicates the highest usage to become the overall Capacity Level that is used to select which constraints apply to batch workloads.
When defining Service Groups for General Services, and Production Service Groups, you can optionally enable ACM constraints for each one and define which constraints to apply at each Capacity Level. For example, you might restrict a given Service Group to 10 active jobs at Capacity Level 4, but only 8 at Capacity Level 3.
ACM can also automatically change the WLM Service Class for selected batch workloads as the Capacity Level changes. In the SLM Policy, you can define the mapping of Capacity Levels to Service Classes. The Service Classes should be members of unique WLM Resource Groups which have been defined with ever smaller maximum values.
There may be times when you wish to set a minimum value for Capacity Level so that some level of constraints is in effect. This can be useful if you have a regular surge of demand at a predetermined time and you wish to make sure that batch workloads are constrained when this occurs. For example, in the brokerage industry there is a large jump in CPU requirements at stock market open.
You can use the SLM SELECT operator command to set a minimum Capacity Level on the system where the command is issued. For example, if you set the minimum Capacity Level to 3, then the Capacity Level will never be 4 or 5. However, if ACM calculates higher Capacity Levels due to threshold values for the enabled Limit Types, the Capacity Level could become 1 or 2 which would constrain the batch workload further.
The minimum Capacity Level will stay in place until reset by another SLM SELECT command or a system IPL occurs.