ILTE abends and performance degradation after change to front-end/back-end

Products

• IDMS
• IDMS - Database

Issue/Introduction

Initially there was a single production CV where all users signed on and all databases where accessed in the same address space. Changes were made to multiple "DC-only" front-end CVs and one back-end CV where all application database access was done. The front-end CVs connected to the back-end by means of SVC type NODE definitions. Once a full load of users was engaged, many "bottlenecks" were seen causing poor response time and also a large number of ILTE abends on back-end CV.

Environment

Release: All supported releases.

Resolution

The SVC type NODE connections to back-end CVs require the use of an ERE (External Request Element) on the back-end. Prior to the switch to the front-end/back-end configuration, there were only a small number of external programs (ERUS) accessing the CV so the sysgen defined only 10 EREs (MAXIMUM ERUS IS 10 on the Sysgen SYSTEM statement). This was the cause of the bottlenecks.

Changing MAX ERUS to 225 on the back-end CV resolved the bottlenecks.

The ILTE abends on the backend were caused by the fact that RESOURCE TIMEOUT was the same on both front-end and back-end. The front-end CVs were changed to have RESOURCE TIMEOUT INTERVAL 240 and the back-end CVs RESOURCE TIMEOUT INTERVAL 300.

This stopped the ILTE abends.