Hi, Here is some nice follow-up information from OTN Database Forum ( OTN Database forum ). This update should give you an idea of not only how columnar data store works but maybe some tips on how to use it as well. The in-memory feature copies data from tables and keeps them in memory in a separate section of the SGA (see prior blogs for syntax and examples). This copied data should include just the columns needed because the structures are converted to this ‘column store’ storage area using vectoring for tracking where the data is when requested by SQL. The optimizer decides if the column store or accessing the underlying table is the best performance method. Oracle doesn’t use the undo or redo features to maintain this replicated data because it’s never saved to disk. Oracle refreshes the column store as the underlying table is updated/maintained. My interpretation of this is to review reporting needs that search large volumns of related data, see which columns are needed, and see if this feature works for you. Remember, this is a separately licensed option from Oracle. Dan Hotka Oracle ACE Director Author/Instructor/CEO
↧