Re: DB2 HADB

"Mark Townsend" <markbtownsend_at_comcast.net> wrote in message news:JLednZ2op-GBkRXeRVn-vw_at_comcast.com...
> Given that you won't discuss what the difference between DB2's table
> clustering and Oracle's table clustering then we will just have to agree
> to disagree then.
>
> For the innocent bystanders, I do refer them to a discussion on the
> differences between IOT's and single table clusters in Oracle, and when
> one would use one of the other - http://tinyurl.com/dc4jl

I didn't say I wouldn't discuss it. Oracle single table b-tree clustering is similar to DB2's MDC (multi-dimensional clustering) whereby a block is reserved for a given key value. This works fine for clustering an employee table by department where there is roughly an equal number of employees in each department. However, it works poorly for last name clustering where you may 200 people named Smith and only one Rielau.

IMO, the closest thing to regular DB2 clustering is Oracle IOT, but obviously there are differences. Namely, IOT's are always in exact order (like a DB2 or Oracle index) and DB2 clustered tables are in approximate order, provided there is room on a page to insert new rows.. In any event, neither Oracle b-tree clustering nor IOT's need to be reorged nearly as frequently as DB2 regular clustered tables.

Regular table clustering in DB2 is used much more frequently than clustering in Oracle, which is why (in answer to Noon's question) DB2 DBA's are more concerned about reorgs than most Oracle DBA's. Received on Sat Nov 26 2005 - 02:15:25 CST