Re: DB2 HADB

Mark A apparently said,on my timestamp of 26/11/2005 12:00 PM:

> does not have. But Oracle does not do clustering of tables like DB2 does it,
> and I think there are some advantages to the way DB2 does clustering.

I'm not so sure about those advantages. Basically, it's good old vsam multi-key indexing. Used to be called MSAM by Univac, for example. (Christ, that's how old the technique is!)

The disadvantage of course is the overflow (stored as you say somewhere else). Which requires periodic reorgs IF the table is accessed by sequence of key as a significant portion of its use.

Mind you, Oracle can be pretty nasty with their row-chaining and row-relocation (essentially another manifestation of a similar problem). I believe there is something now in 10gr2 to address this problem. Online as well, ie: no downtime. I wonder how well that will work with ASSM...

But from what I see the reorg in db2 is a bit of a problem given there are warnings and limitations about it all over the place. Maybe IBM should add a note to the doco that reorgs are only needed WHEN using clustered tables?

Because to someone on the outside it can look like it's an essential part of life, if they don't also read the specifics of table clustering. When like you said: it doesn't need to be.

-- 
Cheers
Nuno Souto
in sunny Sydney, Australia
wizofoz2k_at_yahoo.com.au.nospam