Paging enables the efficient utilization of a relatively small main memory as compared to the number of virtual address space requests.
Thanks to the availability of main memory at a reasonable cost, there is generally sufficient main memory to ensure that dynamic paging does not represent a problem. What follows below therefore only needs to be taken into account in special cases.
Paging is the result of a lack of memory. It burdens both the peripherals and processor. For this reason the paging rates (page reads per second) shown in section "Standard values for BS2000 servers" should not be exceeded. Violating the maximum recommended values is only permitted if the response times are acceptable and if all other workloads (CPU, channels, volumes) are low.
In the event of bottlenecks on the channel or paging volumes, it may be the case that the specified paging rates have not been reached, yet the response times are still unsatisfactory. This is caused by excessively long waiting times for PAGE READ. Poor configuration of the paging peripherals has serious detrimental consequences and must be avoided at all costs.
TP mode is very sensitive as far as paging is concerned - and even more so, the higher the number of important operations performed by relatively few central tasks. Such central tasks must not be burdened with paging I/O operations, since the resulting delays would be passed on to many other tasks. Paging tends to increase at breakneck speed. For this reason it is not advisable to exploit the main memory of a system to its full extent (see also the notes on the “Main memory utilization” report of the report group WORKING-SET as well as the sections "Monitoring main memory utilization" (Characteristic main memory values) and "Paging").