"Two out of three ain't bad"
Have you ever seen a triangle like this?
Although this triangle has multiple names (triple constraint, iron triangle, and project triangle) and multiple variations, the same rules apply: “pick any two.” Choose Good and Cheap, for example, and it will take more time to complete. Choose Good and Fast and it will be expensive. Choose Fast and Cheap, well, we unfortunately all know this one—you get what you pay for with low quality results.
Although this triangle has withstood the test of time since its origin in the 1950s, I have managed to prove it wrong on our current project—sometimes, you don’t even get to pick one of these!
领英推荐
We are currently redoing our bathrooms, and somebody we used for flooring said he knows tile. By the way, we learned there is a difference between knowing tile and KNOWING tile. It took him several weeks, but he finished two bathrooms and was about to start the third. In one of the bathrooms, the tile lines were not straight. In the other bathroom, we ran the water and it pooled in one of the corners instead of going down the drain. We eventually brought in someone who just does tile and nothing else. He has completed demoing the bathrooms that were not done properly, and has been tiling these last few weeks. If we hired this tile expert right from the beginning, we would have chosen Fast and Good in the pyramid, as this guy is expensive. Instead, the cost to buy tile twice and then demo the bathrooms twice was extremely expensive, took a long time, and in some areas, not as high quality as if the tile expert did it right from the beginning. As a result, we got zero of the three on the triangle.
I think sometimes projects are like this, though. I have worked on many data modeling projects where the starting point was a massive amount of “technical debt”. I love this expression, technical debt. It means, like that picture of me above sitting next to our brand new expensive tile ripped out from our bathroom, that we definitely did not get it right the first time. That mound of tile is our technical debt.
Data modeling requires project teams to think things through before building something, and the chances of getting it right are much higher the more we think things through. I am often asked what the Return on Investment (RoI) is on data modeling. If we compare the data modeling project to our home bathroom renovation, where we probably spend over three times as much in tile and labor over doing it right the first time, I would say data modeling has over a 300% RoI!
Senior Consultant in Data Management
2 年I am always flabbergasted by how many software, critical for their businesses, were developed without a professional data modeler. The first version may work (with many hoops), but there will be so many challenges for interoperability, data analysis and simply keeping up with the business evolution, caused by a badly design data model. As usual, your analogy is both simple and efficient... thanks for your articles.
Chief Data Officer ? Scaleup and Transformation ? Data Management Consultant , Advisor, and Coach ? Grew multimillion-dollar revenues ? Drove multimillion-dollar projects ? Data Architecture and Data Governance innovator
2 年Everyone's analogies to the real world are great. I've used them all; building houses, remodeling kitchens and bathrooms, punching and kicking things. I've still struggled with this problem for 30 years in the software/data modeling realm. I've never really made any progress with 80%+ of project teams. It's too late at that point. Ask anyone at the beginning of software or data platform project...would you build a house without a blueprint? "No way, of course not!" But the word software was coined for a reason...it's "soft" ware. You can't see it or touch it. You can't see the crooked tile lines (spaghetti dependencies, goto statements ?? , persistence stores for every micro service, etc.). I'm told, "We have to be agile and make changes as fast as our product owner changes their mind. It's nothing like building a house." Of course, it's easy to change "soft" ware so why waste time with design efforts? The problem I see is a "wet" ware problem. It's in the lack of data and technology literacy among leadership. That problem started long before Sprint 0. Until you crack the leadership, advocacy, and literacy vacuum I agree with Steve Hoberman, DMC, you don't even get to pick one. #data #dataleaders #dataculture
Hands-On Enterprise Data Architect | Leveraging Data Management experience from defining Data Governance & Strategy to implementing Data functions in making Business Data a valued asset
2 年Great perspective Steve! 6-Sigma has a similar triangle of Cost-Time-Quality and the constraints placed on them. I am trying to remember and may have this incorrect, but I remember using a "Software Quality Equation" of Cost-Time-Resources = Software Quality. As you point out, the data model is one of the critical "blue prints" needed and without it technical debt/costs of all kinds will most likely occur.
As often, a great article, although as a resident Aussie, I stumbled over the use of "demo". Here in Oz, it means demonstration (either the political kind or before potential buyers). I don't think I've ever heard "demo" for "demolition" here in Australia! ?? As (I think) Oscar Wilde said, "We are two nations separated by a common language". As I've learned, you need to clearly define the meaning of the terms you use BEFORE you embark on modelling.
Computer Science Adjunct Lecturer
2 年I also thought of it as the Consultant project triangle (Time, Money & Quality) Pick only two.