Journal Chacham's Journal: Link of the day: on the importance of a good data model 8
Post about a post about a post.
Toon Koppelaars blogged about Robyn Sands blog about Bert Scalzo's recent article "Is Data Modeling Still Relevant?".
Bert's article is excellent. Robyn's post and the following comments are amazing. It's like John Brady worked on our team.
Think about it, read it and you have read a post about a post about a post about a post. Think of the potential!
And absolutely relevant (Score:2)
I'm working my way through a database done by young programmers who claim to be trained in Object Orientation, but not database modeling.
To insert just ONE top-level record (a company in a resource-allocation database) I need to figure out a way to insert three records in three different tables:
1. an object record- which contains a *globalized* serial primary key covering several tables.
2. the company record
3. the main office record (which is *REQUIRED* by a 1:1 foreign key assignment)
Third normal form t
Re: (Score:1)
Ugh.
I feel your pain. :(
The problem is two-fold: (Score:1)
1) We have Extreme Programming, and things like it. Where you do the minimum to accomplish the requirements, as they are *today* -- it's counter to the philosophy to design in things for the future.
2) We don't have DBA's, or people like them. Most places I've worked kept costs down by omitting them. So you get databases with no constraints applied, and programmers wasting time adding code to the application to deal with nonsensical kinds of data that should never be coming back from the backend. IMNotSoHO,
Is this anything like... (Score:2)
being the Dude playing the Dude who was the Dude?
As a serious business user (Score:1)
(well before I became unemployed)
having to ask where and how the database captures these business critical measurements sucks.
We get an original set of numbers daily and at the end of the month a verified statement. And as expected there are differences. This was known, incorporated into the design and failed to produce the correct outcome (how much gas we bought and sold in a month).
I had to go back to the boss and have him ask the consultants - where is the data model so we can verify the numbers, trace
Re: (Score:1)
Sheesh.
Good story though.
I'm currently on a project we're we'll be dealing with thousands of financial statements, processing and verifying them for use within reports and the like. We have to have a history of changes, right from the submitted statement. As this is financial, we have to get it right.
I find it odd and disturbing that there was a financial system without a decent model. Do those who code on the fly really think a model is *never* needed?
Re: (Score:1)
I think they saw the specialist knowledge required to produce reports as a future income stream, hence using the code as the business model.
Paul the boss was very displeased when I pointed there was no document other than code - there were ppt files with boxes and lines pointing from external data sources to database, but no names or fields.
But nothing that could be interpreted as final_daily_value at meter_X is held in table Y and is used by process Z (well it wasn't used in process Z was the problem - in
Re: (Score:1)
I wonder if it would be helpful it there were DBAs/DAs for hire offering Data Modeling services. Turn it into an art form that can be sold as a service. Normally, when a DB is used, there should be a modeler on the team, but that isn't always practical. Perhaps then, for those teams that only have programmers, they can view the Data Model as something that needs to be done by someone other than them, and have the team lead/supervisor see it as something important to the business.