IBM Mainframe Forum

Perhaps a bit of background is in order, I'm a mainframe COBOL programmer with some limited knowledge of assembler. I'm doing a performance assessment for a company which does not have an SMF extract / reporting process. So I'm attemping to modify an assembler program to process the SMF 101 record. I'm using as a starting point this assembler program https://ndtdb2.blogspot.com/2017/10/assembler-to-select-only-101-records.html?m=1&fbclid=IwAR0niwHPfYPfS31ilC9yDicMV97NLnKJV0_wYJg3rcF75Xwu8gt0E1hNbeQ

The question I have is that if the DB2 SMF records have been written in a compressed format, is there a way to decompress them in an assembler program by perhaps calling a service module to decompress the record ?

having to run a stand alone utility to read in a file of compressed SMF 101 records and write out the uncompressed records would double the amount of time / resources needed to do the analysis

any help would be greatly appreciated.

Chuck Haatvedt

Are you certain the records are "compressed?" Many fields in SMF data are either packed decimal dates (X'0121dddF' ddd is day of year) or binary time of day and not word aligned in 1/100th second interval.

Thanks for the reply Steve, what I'm concerned about is the following from the SMF type 101 record. Note the the following reference

THIS RECORD IS COMPRESSED BY CSRCESRV

the following is a code snippet from the record layout for the type 101 records.



If SM101BUF > 0, that would seem to indicated that the record was compressed by CSRCESRV.

then the question is how would I go about uncompressing it ?

Chuck Haatvedt

I found the description of the CSRCESRV services in the z/OS MVS Programming: Assembler Services Reference ABE-HSP manual.
They look pretty straightforward to me. I don't have access to a system with DB2 running, so cannot test reading a type 101 - compressed or not.