Improving FEPI performance

You cannot tune FEPI itself because it is already optimized for speed of response. However, you can manage the performance of FEPI applications.

FEPI runs under a separate CICS® task control block (TCB) and CICS permits only one application program to issue a FEPI command at a time. This is a major influence on FEPI performance. Although many application programs can have FEPI commands being processed at any time, only one application can issue a FEPI command.

In a lightly loaded system, CICS does not run FEPI until a command is issued. Therefore, performance is impacted by the overhead of starting up the TCB so that the FEPI command can be processed. In a heavily loaded system, this overhead is not present, because the TCB is already active processing earlier FEPI commands.

FEPI tries to minimize this overhead by issuing timer requests that ensure that the TCB is not inactive for more than one second.

There are three main principles that can be used in FEPI applications to provide the best performance:
  • Minimize the number of commands. Each FEPI command generates a CICS WAIT, even if no network transmission is involved.
  • Keep data transmission to a minimum.
  • Avoid disconnecting sessions.

Application development techniques are described in FEPI application techniques for performance.

For FEPI system programming, you can reduce the number of commands by using lists of resources on a command where possible. However, if you use a list that results in a z/OS® Communications Server operation, the following conditions might occur:
  • Flood z/OS Communications Server by requesting too many operations at once
  • Flood the back-end system with requests for session initiation
  • Flood the front-end system with started begin- or end-session transactions.
So you must carefully evaluate the benefits of using lists.


dfhp71x.html | Timestamp icon Last updated: Monday, 5 June 2017