2. PERFORMANCE REPORTING ANALYSIS › 2.4 I/O Configuration Analysis › 2.4.8 FICON Channel Performance Analysis Inquiry

2.4.8 FICON Channel Performance Analysis Inquiry

The introduction of FICON (fiber optic connectivity) channels
into the mainframe platform has helped resolve I/O
performance problems by:

- improving the path speed and throughput
- simplifying I/O configuration design by reducing the number
  of channels needed
- permitting installations to locate devices at remote sites,
  up to 100km away.

To manage these new channels, IBM has introduced several
performance metrics into the SMF type 73 record, which are
included in the HARPCA file.

To gain a complete picture of FICON performance, you must be
able to view performance in the same manner as the channel
itself and to relate those measurements to services times.
This requires developing estimates of two new FICON specific
metrics: Open Exchanges and Credits.

Open Exchanges (or OEs for short) may be thought of a as an
MPL for the channel; that is, a measure of concurrency.
Credits are a hardware entity, similar to a buffer that
contain both the channel program and data for the in-flight
I/O operation.

There are specific hardware limits placed on both Open
Exchanges and Credits, and exhausting either of these
resources will result in channel saturation and poor
performance.

This query is designed to compute these new measurements and
summarize the data according to the user's specifications and
to then produce a series of graphs and reports, depending on
execution options.

The following sections describe the facility's operation and
reports:

    1 - Standard Output
    2 - Technique Tutorial
    3 - Case Study
    4 - Component Operation