My recollection is that even the original IEBCOPY was faster than IEHMOVE. Of course, it only supported copying between pre-allocated PDSs.
-- Shmuel (Seymour J.) Metz http://mason.gmu.edu/~smetz3 ________________________________________ From: IBM Mainframe Discussion List [[email protected]] on behalf of Mike Kerford-Byrnes [[email protected]] Sent: Thursday, December 22, 2022 5:07 AM To: [email protected] Subject: RLSE - PDS compression etc. Back in the time......... The recent posts re the above brought back some old memories. These date back to the late 60s early 70s when OS MFT/MVT reigned supreme. At that time there was no fast IEBCOPY and the "traditional" way of reorganising libraries was to use IEHMOVE (Unload/reload). This took forever and the Control card syntax often confused the unwary.. Then I heard of a program written by an IBM SE called Nick Tindall, based in Bristol, UK. It was called SUPRSHFL and it did in-place compression of Load module PDSs at an incredible speed - at least for the time. As compression was in-place, there was a risk, so Kodak (my then employers) integrated it with the nightly backups so as to have an up-to-date image just in case. It only worked on 2314 drives. 3330s had just been announced so while it was of immediate use inn its current state, it could not be viewed as a viable production feature. In February 1972, I was assigned to a Month long project at Kodak head office in Rochester, NY researching the 3330 and preparing the plans for their implementation at various company locations. Before I flew over, I contacted Nick Tindall and met with him to see if 3330 support was forthcoming. Not from him, he was assigned elsewhere. But he gave me a lot of pointers and some microfiche of some of the assemblies (Not the ones that mattered, as it happened). It used EXCP with heavily chained Read Count Multitrack to create CCW chains of READ CKD each of which read a whole track (This was way before the Read Full Track CCW) . The directory was read into storage and as each "data" track was read into memory, the blocks were "shuffled up together" and the in-core directory updated. The "data" tracks were then rewritten and finally the updated directory was restored.Note lists and other TTR related stuff was also covered. So armed with (IEHMOVE) unloaded tapes, fiche and time (February in Rochester is not conducive to going out in the evenings), I set out to derive the changes to make it work on 3330 (quite a lot of ZAPs). After a number (undisclosed) of attempts, I managed to get it working, and subsequently passed the modifications back to Nick Tindall. It is notable for me since it introduced me to the world of EXCP from which a number of other projects evolved. I used the concepts to write a high-speed unload/reload of PDSs without reorg - just to avoid the interminable IEHMOVE that was all that was on offer (IEBCOPY did not support tapes then) and some high speed full track unload/reload of any format of dataset over a file transfer product. I lost contact with Nick Tindall after that - that was a shame - he was a VERY smart person. I have fond memories of those times - the events subsequently proved to be a significant part of the foundations of my career in IBM mainframe computing, - so I am eternally grateful. Merry Christmas everyone and a Very happy New Year to you all Mike Kerford-Byrnes ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
