January 4, 2012 at 5:14 am
EdVassie (1/4/2012)
Most SQL code can be written by anyone in the sprint team.
Gosh... I've worked in shops that have taken that stance. Unless they generally keep things to simple C.R.U.D., it usually turns out to be a performance disaster down the road (and not very far down the road, at that).
--Jeff Moden
Change is inevitable... Change for the better is not.
January 4, 2012 at 5:17 am
Dev (1/3/2012)
Jeff Moden (1/3/2012)
MysteryJimbo (1/3/2012)
Jeff Moden (1/3/2012)
There's also a matter of timing here. I've found that if you start the database development before you start any of the front end development, you end up with a much more correctly designed database that works well and tends to be much more scalable.Very true. At least 5 days head start depending on the upfront requirements already being gathered. In an agile way, the requirements can then be expanded and added to during development based on a good upfront design.
In some companies I've found the bad practice opposite. Some tables have been directly linked to the fields on a form rather than correctly modelled. If they were on a single form, they went into one table.
Agreed... a single table for each screen where all integer columns were BIGINT, all character based columns were NVARCHAR(4000) or NVARCHAR(8000), and little, if any, concept of what normalization is (multiple phone numbers and addresses in the Customer table, for example).
Yup... comma separated or pipe separated columns. And within few days of development / maintenance they will come back to SSC & ask βI have CSV columns, please help me in getting in a tableβ.
Indirectly, we will sell String Splitter. π
Those are always fun. An OTLT (One True Lookup Table in the form of an EAV) is much more fun! After all, you only need one index for such a thing. π
--Jeff Moden
Change is inevitable... Change for the better is not.
Viewing 2 posts - 16 through 18 (of 18 total)
You must be logged in to reply to this topic. Login to reply