Re: using temp tables for staging databases?

[K Gopalakrishnan]

Hi,   The walking in the freelist is just 5 blocks (or the value of _walk_insert_threshold number of blocks) and  I belive tanel is talking about (_release_insert_threshold) unlinking from freelist, which also default to 5 blocks.   KG ----- Original Message ----- From: Binley Lim To: Multiple recipients of list ORACLE-L Sent: Monday, October 27, 2003 4:59 PM Subject: Re: using temp tables for staging databases? What you have described is expected behaviour - if "the next insert would drop amount of free space less than PCTFREE", the block is unlinked.   What I had not considered is what happens when the block is still below PCTUSED? As usual, Steve Adams' website explains this very nicely - it is not unlinked, just walked over, so the next smaller insert is likely to succeed. And as you pointed out, It would indeed take rows that are a large proportion of blocksize for this effect to have an impact. ----- Original Message ----- From: Tanel Poder To: Multiple recipients of list ORACLE-L Sent: Saturday, October 25, 2003 10:09 PM Subject: Re: using temp tables for staging databases? No, blocks won't fill up to PCTFREE in case a block is already above PCTUSED and the next insert would drop amount of free space less than PCTFREE. Blocks are just unlinked from freelist in this case. That means if you normally have 10byte inserts and occasionally have 4000 byte inserts in your table, then you might be wasting space due to premature unlink of blocks in freelist.