Re: Oracle 8 vs. 8i for DW -- opinions?

You might also mention Summary Management - which allows Oracle to rewrite a query against a large table to use a previously declared summary table - on the fly, without the user knowing it's happening.

Hash partitioning and sub-partitioning which can make it easier to balance I/O loads and make query access times more uniform, as well as allowing a degree of 2-dimensional partitioning.

Partition-wise joins (typically between equi-partitioned tables). Which can have a geometrically staggering effect on performance.

On line rebuilds of tables (IOTs)

Secondary indices on IOTs

Proper temporary tablespaces that do not need to be backed up - and can be rebuilt in fractions of a second.

Locally managed tablespaces to reduce the stress on the data dictionary - particularly useful when running parallel query and index/table builds.

The stored execution plan feature is probably not going to be useful, as it depends on exact text match - it's really aimed at pre-programmed OLTP systems with application servers supplying the same SQL time after time.

etc. etc. etc.....

Refers to the change as an UPGRADE, rather than migration, and it consists largely of installing the software in a new Oracle_home - fixing up the init{sid}.ora, and starting up and running 2 or 3 SQL scripts. There are a couple of restrictions about which versions you can upgrade from though (8.0.2 and 8.0.4S are out, but 8.0.3, 8.0.4 and 8.0.5 are okay).

--

Jonathan Lewis
Yet another Oracle-related web site:  http://www.jlcomp.demon.co.uk

PerlMongerWannaBe wrote in message <7thcm0$3iq$1_at_nnrp1.deja.com>...


>Hi erik.


>

>In terms of warehousing in 8i, there are some new features but I'm not

>sure they are worth the translation from 8.

>

>The significant one's from a warehousing perspective would be:

>

>Function based indexes (would be big for ad-hoc queries)

>Ability to store execution plans after statistics generation

>The new SORT_MULTIBLOCK_READ_COUNT parameter