If the service times or the percentage of requests changed exceed the recommended values, there are many potential sources of delay you can take a look at. In general, any configuration or workload changes that affect the observed CF request service times, or result in a change to the calculation of the conversion threshold, may result in substantial changes in service time. Examples of such changes are listed as follows. In subsequent sections under this topic, the indicators of some problems and corrective options are discussed.

• Insufficient CF capacity
• Contention for CF paths
• IOP contention
• Shortage of CF subchannels
• Significant increase of CF subchannels.
• Changing to a faster or slower type of CF link technology.
• Upgrading or downgrading a z/OS® processor to a faster or slower technology, or changing the number of online processors in a z/OS image (which changes the effective processor speed somewhat).
• Upgrading or downgrading a CF processor to a faster or slower technology.
• Adding or removing CF processors from a CF image.
• Changing the shared/dedicated nature of the CF processors, or, for a CF with shared processors, changing the LPAR weight of the CF partition, or enabling/disabling the use of Dynamic CF Dispatching.
• Changes in the amount of distance latency that occurs between the sending z/OS and the target CF.
• Changes in the CF Level of the target CF.
• Changes in CF utilization, caused by workload changes or other factors (such as placement of additional structures in a particular CF).