Hi DB2 experts, have some questions on combining compression and replication. We are evaluating the risk turning on db2 compression on capture side.
From the KC article (ref. link), sounds like there's the risk that db2ReadLog API can't decompress a log record, when the table-level dictionary is lost. Is there any best practice we could follow to avoid getting caught in such situation? It could very well be "don't use compression on capture side"......
We don't use LOAD or REORG, and understand there are options to keep the old dictionary, but haven't found definitive answers fro KC, about how to keep the old dictionary for some admin operations (eg. admin_move_table, add column to compressed table). We could test things out by ourselves (SYSPROC.ADMIN_GET_TAB_DICTIONARY_INFO seems like a good option), but just wondering if there are already some references we could use.
Also we plan to use adaptive compression, which updates page-level dictionary automatically. From our understanding, page-level dictionary shouldn't matter, since it won't effect the contents in transaction log. Is that correct impression, or should we actually worry about page-level dictionary being lost as well?
Last but not least, we use LUW 10.5 FP8.
Thanks for your help.
Rui Chen[Organization Members] @ May 24, 2018 - 05:47 PM (America/Eastern)
Rui Chen[Organization Members] @ May 24, 2018 - 05:49 PM (America/Eastern)