Re: how to place a table in memory

>> Actually, if there are blocks that are cached, they will slow down
reading the table via a full table scan, as such blocks will cause less than full extents to be read <<

I'm intrigued by this comment. The scenario (correct me if I'm wrong) is that if a few scattered blocks of a table are present in the SGA, then the process of full-scanning the table is disrupted by them? I'm not convinced by this, and I'm trying to think of a way of demonstrating or disproving it.

I can see that if some of the table's blocks in the SGA were dirty then those blocks ought not to be over-written by the clean blocks being read from disk, but whether this would make the actual multiblock read operation slower, or whether it would prevent multiblock reads in some situations, I'm not sure.

Any other opinions? Received on Fri Feb 11 2005 - 17:04:46 CST