Niall Litchfield wrote:
> <mikharakiri_nospaum_at_yahoo.com> wrote in message
> news:1112914442.722412.78720_at_g14g2000cwa.googlegroups.com...
>
>>fitzjarr..._at_cox.net wrote:
>
>
>>Finally, the truth surfaces. So, unlike the other, really useful
>>features, you can't really defend partitioning ("it's not for
>>everybody","your mileage may vary","no silver bullet").
>
>
> It seems to me that this is rather turning the argument on its head. Earlier
> on in this thread you mentioned two features which you did think were
> useful, materialized views and bitmap indexes.Both of them are useful, but
> neither of them are always useful. The contributors to this thread other
> than you have suggested that having partitioning available was helpful in a
> number of situations. I think that I have seen the following possibilities
> suggested
>
> 1. You regularly delete significant amounts of data based on what would make
> an excellent partition key.
> 2. You regularly load data into a warehouse.
> 3. You wish to reduce the backup resource requirements by making old
> partitions read-only.
>
> In other words just like materialized views or indexes, sometimes the tool
> of partitioning can be a help, other times it may not be.
>
> Incidentally in your last response to the deletion example you suggested
> that
> a) the DBA could choose when to do the delete
> b) users were isolated from the transaction (presumably off the system)
> c) Users don't care about the old data.
>
> That rather adds to the scenario posited, what happens if one or all of
> those further assumptions are incorrect.
Anyone that thinks that any aspect of any product is a "silver bullet"
is engaged in some form of self-delusion.
--
Daniel A. Morgan
University of Washington
damorgan_at_x.washington.edu
(replace 'x' with 'u' to respond)
Received on Sat Apr 09 2005 - 11:02:43 CDT