In general, if you segregate message traffic according to size, signaling performance is improved. This would be true even if there were no additional overhead incurred when XCF had to format larger buffers. If large and small messages are intermixed, the small messages tend to be delayed simply because the longer messages take longer to process.

In the Figure 1, the SML and BIG columns indicate when it may be useful to segregate messages by size. The reports can indicate that such a change might be useful, but not what the appropriate size should be, only that it should be smaller or larger.

Information about message sizes can be determined by creating some temporary transport classes (use the SETXCF operator command) that use different class lengths. Then use the RMF™ reports to investigate the appropriateness of that class length. By assigning a unique group to those temporary classes, you could determine the range of message sizes used by that group. Alternatively, the message sizes used by the group may have been documented and you could use that information. This information might indicate that a particular group is sending the messages that don't match the defined buffer size. That group can be assigned to its own transport class in order to segregate it from the other users of the message service.