Re: Index query - monotonically increasing columns

Howard J. Rogers wrote:

Morning Howard,

> Maybe my other post hasn't shown up yet....
Your other post did show up, but after the one I replied to :o) What hasn't yet shown up is my reply to your other post which I replied to via Google yesterday afternoon. Sigh !

> but what difference is there
> between random letters and reversing the byte order? None is the short
> answer. Both techniques cause the entire base of the index to receive new
> entries. But one relies on developer code. One on Oracle code.
True. I suspect relying on Oracle code may well be the better idea - if I go down this route.

>
> And yes, there is another way. What you're after is a mechanism which,
> whatever the particular values being inserted, ensures the inserts take
> place evenely across a lot of physical locations (thus eliminating
> contention issues). That is a perfect description of hash partitionining.

Which is what I tested out yesterday evening on a pure number column and 'it just works'. The point of this exercise is for me to learn about stuff we don't use at work and to see if it can be used at work. We have a few tables in some apps which are keyed on a sequence number, there are deletions but inserts are always higher numbers. Hence my experimentation and need to know if there were alternative solutions.

It also comes in handy for selfish reasons as well :o)

Thanks again for your answers.

Regards,
Norman (at home).

-- 
Remove a certain other database to reply :o)