Hi community,
I am currently running an Oracle Database RAC environment (version 19.17, 2-node RAC on Linux x86-64, ODA X7-2 HA).
Due to space constraints on our primary ASM diskgroup (+DATA), I am planning an emergency storage tiering operation. My strategy is to move some of the heavy index datafiles (*_idx.dbf) from the +DATA diskgroup over to either the +RECO diskgroup or a mounted NAS storage area to free up space immediately.
I intend to use the online data-pump movement feature introduced:
ALTER DATABASE MOVE DATAFILE '+DATA/orcl/datafile/sbk_idx.dbf' TO '+RECO/orcl/datafile/sbk_idx.dbf';
or
ALTER DATABASE MOVE DATAFILE '+DATA/orcl/datafile/sbk_idx.dbf' TO '/oranas/orcl/datafile/sbk_idx.dbf';
I would like to clarify a couple of physical storage architecture details with the experts here:
- Index Status: Since this is an online movement, will the underlying tables' indexes become
UNUSABLE or corrupted during/after the execution? Will a manual REBUILD INDEX be required once the move completes? (My understanding is that the logical file_id and relative block_id inside the Control File remain unchanged, meaning the ROWID indicators in the index leaf blocks should stay 100% valid, but I want to double-check).
- Data vs. Index: If I have to perform the exact same command on a datafile that contains actual table Data instead of indexes, will there be any architectural differences or hidden operational risks besides potential network I/O throttling during Full Table Scans?
Any insights, best practices, or official MOS references regarding this online migration would be highly appreciated.
Thank you in advance!