I'm not experienced enough to know exactly how much benefit it would be, but
surely if you have a table on one physical mountpoint and the index to the
table on another physical mountpoint then this would reduce contention.
Obviously, you would seek to reduce I/O hotspots. Just because you have the
option to use a default index tablespace, doesn't mean you have to use it.
Having the option to set a different index tablespace to your default
tablespace just means that you have extra flexibility. Oracle obviously has
a great deal of flexibility currently. Having your indexes in different
tablespaces also allows you greater flexibility when it comes to archiving
as well. If disk space is a consideration then why bother archiving an index
when it is redundant information? Simply being able to drop the indexes and
then the tablespace means greater flexibility. Over half of one our
databases is taken up in indexes, this means a lot of time is spent
archiving indexes when they could simply be dropped and then archived. The
time used to rebuild the index isn't an issue if the database had to be
restored.
"Howard J. Rogers" <howardjr20002_at_yahoo.com.au> wrote in message
news:OCMka.10144$1s1.168134_at_newsfeeds.bigpond.com...
>
> "Peter" <no_email_at_no_email.com> wrote in message
> news:b701am$bhc$1_at_bunyip.cc.uq.edu.au...
> > Cool, I was wondering when they would bring this feature in. Thanks for
> the
> > info! I just hope that it's not a violation of Oracle's security model
:)
> >
> > Now we just need to be able to make a default index tablespace for the
> > users.
> >
>
> With a certain sinking feeling, can I ask why?
>
> There is no need for indexes intrinsically to be housed in their own
> tablespace, but rather you should seek to eliminate I/O hot-spots and
> sources of I/O contention, which can arise as much from an index-index or
> table-table bun-fight as from an index-table one. Therefore, such a
default
> index tablespace would be pointless, erroneous and deeply misleading.
>
> I'm glad we haven't got to that stage yet.
>
> Regards
> HJR
>
>
Received on Tue Apr 08 2003 - 23:15:15 CDT
