Dear friends,

I am developing a db which appears to have a potential bottleneck. Every person would need to access and update an indexed view based on one table. This table has is star joined to a heirachy of parent-child descriptive lookup tables. Do I de-normalise by adding a set amount of replicate tables to alleviate the bottle neck. This can work as far as the business logic. I just dont know if the db will run effectively. Please advise on standard denormalisation techniques.

OxyJen

rule #1 for denormalization: don't denormalize until you have an actual performance problem, not a "potential" problem

thankyou very much

OxyJen

this table sounds like a fact table joined by a a bunch of dimension tables. if that is true, im going to have to disagree with r937.

#1 rule on datawarehouse: ALWAYS denormalize.

if, however, it is just a "simple db" or an OLTP, then, by all means do not denormalize.

HTH

well spotted, duul, i missed the word "star" somehow

Thanks Guys,

I am afraid that I am a newbie and I do have star shaped detail tables with nested lookup tables. But they are not dim tables. I am using OLTP cos my clients will need to frequently update their records.

I have reworked some of the tables so that the schema are trim.