November 14, 2025 at 12:00 am
Comments posted to this topic are about the item Being Mindful of Design Time
November 14, 2025 at 4:15 pm
It will be interesting to see how many comments are made on this topic of database design. It seems that action is what everyone wants to see and design is not action. Like driving to work, some people will go out of their way and take longer to get to work just to keep moving and not sit in traffic. Database design is like planning the trip as opposed to flying by the seat of the pants.
Few managers understand the importance of good design which is based on good requirements. No one wants to take the time to gather the requirements and design the conceptual database before bringing on developers. Many think it is faster to bring on the full team and start designing as you code. This results in a sub-optimal design that performs poorly, requires a lot of computing resources and is difficult to expand, just to name three problems.
Taking the time to design code is just as important but most developers just start writing lines of code with no design. I would ask the question, how many coders actually do any kind of design on paper before writing code?
November 14, 2025 at 6:50 pm
Back in 2014 I went to QCon in London. MongoDb (the 10Gen) had a presenter who said the data modelling is MORE important in the NOSQL world, not less. He was and is absolutely right.
A good physical data model does more than store and retrieve data at pace; it communicates with its users. If it doesn't communicate, then it is a failed model. Just because a document or key-value database does not have a schema does not mean that there is no model for the data. The problem with anything that does not enforce discipline is that the discipline fades and is eventually forgotten.
Perhaps it's just my perception, but data quality seems to be worse than ever. If I might liken data quality to a bar, it has always been like the bar on a hurdle rather than the bar on a pole vault. Now it's like the foot rail around a bar.
November 14, 2025 at 8:41 pm
Agreed, we get more data, but lots of it suspect. Maybe AI will help clean some up, or at least get us closer to an MDM world, but design is low on people's radar (or desire)
November 15, 2025 at 1:01 am
I don't believe AI is going to clean up bad data. Humans can't even clean up dirty data. I have seen situations where bad code created 750 duplicate entries for a person but no one noticed until one day end users were updating the data and multiple rows were returned. Going forward the end users had to use the surrogate key to make sure they had the correct person whenever they needed to update any associated data. They wouldn't clean up the data because they were afraid there may be a different person with the same name with different information. The design error was that they did not think a natural key needed to be identified, just surrogate keys. There were a lot of duplications that could not be cleaned up. Garbage In Garbage Forever
I think design is dead and there will be no good designers or designs that an AI could even learn from.
Viewing 5 posts - 1 through 5 (of 5 total)
You must be logged in to reply to this topic. Login to reply