Bug 31278280 - Degraded Performance with Oracle 10g Client Pro*PL/I Application after upgrading to z/OS 2.2
(Doc ID 2674835.1)
Last updated on APRIL 17, 2023
Applies to:Oracle Database - Enterprise Edition - Version 10.2.0.2 and later
IBM z/OS on System z
When an Oracle 10.2.0.2 Client Pro*PL/I application (PGM=DNIVEND) on z/OS 1.13 (GA 2011) query's Oracle 22.214.171.124 server on Linux x86-64, rows are returned in seconds, and then the Pro*PL/I application loads the data into a VSAM dataset.
After upgrading to z/OS 2.2 (GA 2015), the same set of query's from Pro*PL/I application now takes over 3 minutes and network trace shows it is only
sending one row at a time with smaller sized packets.
They ran a network trace of their Pro*PL/I application running on LPAR1 with z/OS 1.13 which does not show any latency problem contained in file
"lpar1_pcap_trace.cap", and also ran a separate network trace on LPAR5 with z/OS 2.2 using the same set of query's where only one row at a time is sent and is contained in "lpar5_pcap_trace.cap".
Degraded performance was noticed even before precompiler and Binder was run for the Pro*PL/I application (PGM=DNIVEND) where upgrading z/OS from 1.13 to 2.2 was the only change and job DNIVEND ran for 10 hours in order to query a million rows which normally runs in less than an hour.
Then Pro*PL/I application (PGM=DNIVEND) was modified to only query 3393 rows and is when precompiler and Binder was run.
Customer has other z/OS applications that query Oracle using 10g Client on z/OS and one reason they think this is not occurring may be
because those other jobs use arrays and query fewer rows.
When customer ran the same set of query's using PGM=SQLPLUS batch job, it returns 3999 rows in 12.61 seconds on
LPAR1 (z/OS 1.13) and in 12.62 seconds on LPAR5 (z/OS 2.2) so there is no degraded performance with SQLPLUS.
To view full details, sign in with your My Oracle Support account.
Don't have a My Oracle Support account? Click to get started!
In this Document