| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Usenet -> c.d.o.server -> Re: Large initial extent after import
On Sat, 14 Nov 1998 18:07:27 GMT, pantah_at_commonsense.a2000.nl (Jan
Dolman) wrote:
>I've been thinking about this problem some more but I still don't
>_understand_ the implications. Apparently, the number of blocks will
>be the same no matter what size of extent you choose (i.e. you can
>create 11.000 extents of 6 blocks or 6 extents of 11.000 blocks). So
>you basically choose to:
>a) keep the number of blocks (per extent) down, or
>b) keep the number of extents down.
>
>Can anybody _explain_ which option is better, I mean, it is pretty
>easy to understand that chained rows will have impact on performance,
>but I still don't see any reason to choose either option (apart from
>the performance gain Neil noticed, well, can anybody explain that
>one?).
Multiple extents per table can *theoreticaly* have impact only when you are performing table scans, ie when you are not using indexes. However, if the extent sizes are chosen correctly in regard to DB_BLOCK_SIZE and DB_FILE_MULTIBLOCK_READ_COUNT init parameters, then you can neglect the impact of multiple extents on performance even when performing table scans.
If you want to read more technical in-depth explanation of this fact
take a look at *excelent* paper "Avoiding A Database Reorganization"
by Craig A. Schallahamer
(http://www.europa.com/~orapub/papers/pmain.htm).
>Thanks
>--
>Jan
>pantah_at_commonsense.a2000.nl
>(replace "commonsense" by "speed" to reply)
HTH,
Jurij Modic <jmodic_at_src.si>
Certified Oracle7 DBA (OCP)
 |
 |