Ralph Kimball

You go like, huh.

And it's a, it's, but I feel like, There's going to be more of a need for highly curated, well-done books, say, in our industry, for example, right? Because on the level of like what Ralph Kimball had produced back in the day or, you know, versus just slop books, which you can find on Amazon, the number of books being published is ironically going up while reading rates going down.

But It's because anyone can make a book, so to speak.

And even though they probably didn't read it themselves.

So dimensions and facts come from a modeling methodology called dimensional modeling.

which has been around for a long time and which in my opinion is sort of still the best way to be able to provide structured information for reporting purposes, right? So a gentleman called Ralph Kimball came up with this idea back in the 80s and I still think it's relevant today.

So best way I can describe a dimension and effect is if you think of a pivot table, You've got that little quadrant down on the bottom right.

Anything in the right-hand bottom corner, a value, that's what goes into a fact table.

So you almost use the AI as an impromptu teacher.

And then at the end of it, you can say to it, all right, well, based on what you've seen, help me prepare a plan that I can work through to learn more about the particular topic.

and then you just get and as a tip get it to write it to a file so you can create a folder that's called my my dimensional modeling course and have a to-do list and a context and a whatever and then you can get the ai to remember that so that if you if you if the prompt cache gets wiped you've got your learning plan there essentially so in my case i was fortunate enough that i worked with a consultant who gave me the basics of it and then at the end of 2009 i think it was ralph kimball did one of his last dimensional modeling courses in south africa so i was lucky enough to attend that and also then the next year was the etl in depth and then from there i'd leverage that that training to go and build probably three or four data warehouses on top of that capability And then when I moved to Australia after I left PwC and I was looking at the state of where I'm at at the moment, I was introduced to this concept called Data Vault, which is like the third major modeling methodology that I've kind of looked at.

So third normal form, as I said, is very good at reads and writes, the whole insert, update, delete kind of Not bulk things, but rapid transactions.

So I, most of my thirty years career, I spent a lot of time in data.

So I worked with Mr. Ralph Kimball, who's one of the forefathers of data warehousing, and worked with a lot of midsize and enterprise manufacturing healthcare companies.

One of the challenge, and I'm still able to convince it, but I don't have the right formula, so I keep asking everyone, every tech leader that I talk to, it's very difficult to sell or convince, however you wanna put it, data warehouse projects.

We don't use that word to them, but since we are both understand it clearly, so I'm using those words.

I even think with all the talk about context layers, I think that eventually that's where you have to be in order to have NLP, right? The reality is in order to really have that kind of natural language processing where you have an exec that's just going to a chat bot and asking a question to get an answer, then that's where you have to get to.

want to unpick that a little bit more then so yeah i'm a big fan of business of analysis and modeling from lawrence core right it's kind of how i started capturing data requirements many many years ago and lawrence is a hardcore dimensional modeler right he worked with ralph kimball he's like he is that is the data modeling technique that everybody should use and so half his book is around how to capture data requirements using beam and half the book is how to translate those requirements into dimensional models and deal with all the edge cases Um, when I got to the physical modeling stuff, I kind of became more of a data bolt person by default.

Yeah.

And actually I'm just reading Joe Reese's, uh, mixed model arts book right now.