Re: data modeling conundrum

Hi,

Like the last post, I'm working on assumptions here too regarding the nature of the business you're trying to model. You indicate that you never have to query the number pairs. It also sounds like the "pair string" is not growing over time which would allow your column to become indefinitely wide.

So, in deciding which way to go I would consider this; if a given set of number pairs is functionally dependent on the parent record id, so that specifying a parent id 'forces' one given set of number pairs, then include the number pairs (even as one long string if that's ok for whatever you're doing) as a record in the parent table. I.e. 1 parent id equates to 1 set of number pairs.

In contrast, if it's more accurate to consider one parent id as having 'many' number pairs associated with it (the way many sales calls can be associated with a customer) go with the child table for your number pairs.

Phew! I hope some of this made some sense! If not, feel free to email me .
Steve

In article <8v1mvt$5o7$1_at_nnrp1.deja.com>,   bdeneaud_at_my-deja.com wrote:
> I need to store large sets of number pairs like 1.01, 0.01; 1.02,
 0.02;
> etc... in my database. I need to store about 400 of these pairs for
> each parent record. (parent records growing at 100,000 rows a year)
> How can I store these efficiently and what is the best datatype to
 use.
> Would it make sense to store the 400 pairs all as one record in a big
> long string
> of numbers like a varchar(4000)?? or should I create a new record for
> each pair (which seems like a bad idea, because my table would get
 very
> tall and I don't need to search on these number pairs.
> Or are there other better alternatives.
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
>

Sent via Deja.com http://www.deja.com/
Before you buy. Received on Fri Nov 17 2000 - 07:47:20 CST