Theo, We have been using RLS extensively since 2001. Hit many ussues [like skipping records at a CI boundary, getting inapplicable VSAM response codes, etc., see related APARs] most of which are resloved by now. But even last year we had a production problem with a CICS RLS scanner receiving a premature EOF response [resolved by closing, copying and re-opening the file]. We also had several more recent cases where an RLS file closed to ALL online regions lost some trailing records while being copied by a batch job.
However, my biggest problem with RLS is the current implementation of the NOSUSPEND option - it waits for at least 1 second before CICS returns the RECORDBUSY response. With our current transaction volume and processing speed "at least 1 second" is eternity, so instead of using the built-in record-level-sharing feature our applications now wrap all RLS I/O in a pair of Global ENQ/DEQ which allows to perform UNCOMMITTED reads without loosing read integrity. Ugly, but serves the purpose. Overall I woud say RLS is great step forward from the previous "CI-level-locking" version, just make sure you stress-test all your scenarios. HTH, -Victor- ============================================================================================ Hi All, In our shop the VSAM RLS functionality was enabled. SMSVSAM is running with only DFHSM using this functionality. I need to convince the application and CICS owners to consider using the RLS for their VSAM files. Those who are already running with RLS for a while, what improvements have you noticed, and what would you say the real benefits are? Also what problems have you encountered with RLS. I would appreciate your input here. Regards Theo ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
