Joe, I go back to 1997 of the MDS stack I used (in dinosaur terms the Triassic Period 😆). There are fundamentals in any space that are always there. The “how” may keep improving and determining what is “marketecture” (aka recycled terms or new terms that really mean the same thing in the hopes of selling something) and what is not is the constant thing we are always evaluating. Is it just chasing the “the shiny new thing” which is a distraction or does it make us evaluate and research it for integration and improvement. I have tried to use the approach of making a MDS or complete solution set broken down into patterns within the MDS component stack that are based on describing them as a “What” verses a “How”. From there each component is given a definition and an owner to manage it within MDS. From Researching what is “the latest and greatest” to filtering out hype and feasibility to Proof of Concept to Purchasing and Integration into the MDS and pattern(s). I have found that having some type of disciplined approach helps navigate the present environment you are working in. It also, can at times, remove some of the emotion on this capability verses that capability. They all can be an option until ruled out based on using this approach I described above.
"If you disagree, please travel with me in a time machine to 2011, and let’s negotiate an on-prem data warehouse and ETL tooling contract with a giant tech vendor of that era. Otherwise, read on."
Could you please share your ideas about Azure Data Factory? It is similar to on-prem ETL tools.
AWS and GCP prefers modern data stack. Developers need to develop code for implementing data pipelines in AWS and GCP. I do not know any advanced ETL tool in these public clouds.
On the other hand, Azure prefers gui based tools like Azure Data Factory and Synapse Pipelines.
I wonder why AWS and GCP do not provide similar/alternative services?
I’ve noticed, no doubt because of this discussion over the last week, that a lot of vendors are pushing the “modern data warehouse” and it feels just as vapid as the MDS. Like sure, it’s a lakehouse architecture and it’s more modern than your ERP, by decades, but that doesn’t actually solve any problems! It’s just fluff 🪶
Joe, I go back to 1997 of the MDS stack I used (in dinosaur terms the Triassic Period 😆). There are fundamentals in any space that are always there. The “how” may keep improving and determining what is “marketecture” (aka recycled terms or new terms that really mean the same thing in the hopes of selling something) and what is not is the constant thing we are always evaluating. Is it just chasing the “the shiny new thing” which is a distraction or does it make us evaluate and research it for integration and improvement. I have tried to use the approach of making a MDS or complete solution set broken down into patterns within the MDS component stack that are based on describing them as a “What” verses a “How”. From there each component is given a definition and an owner to manage it within MDS. From Researching what is “the latest and greatest” to filtering out hype and feasibility to Proof of Concept to Purchasing and Integration into the MDS and pattern(s). I have found that having some type of disciplined approach helps navigate the present environment you are working in. It also, can at times, remove some of the emotion on this capability verses that capability. They all can be an option until ruled out based on using this approach I described above.
joes ability to tell it like it is, with style, is unmatched!!
Thanks!
In the late 2010s I bought Data Warehousing for Dummies, obviously I was completely oblivious to what was going on in the industry.
The contents of that book is criminally negligent, and I don't think it's been updated since.
I have a question about the below part:
"If you disagree, please travel with me in a time machine to 2011, and let’s negotiate an on-prem data warehouse and ETL tooling contract with a giant tech vendor of that era. Otherwise, read on."
Could you please share your ideas about Azure Data Factory? It is similar to on-prem ETL tools.
It’s decent. depending what you’re doing
AWS and GCP prefers modern data stack. Developers need to develop code for implementing data pipelines in AWS and GCP. I do not know any advanced ETL tool in these public clouds.
On the other hand, Azure prefers gui based tools like Azure Data Factory and Synapse Pipelines.
I wonder why AWS and GCP do not provide similar/alternative services?
AWS Glue has a GUI option. For GCP, it’s more code-based, but that’s fine.
Pertaining to the section mentioning bad practices, I would love to hear your take on modeling approaches which move away somewhat from traditional dimensional modeling, such as this for example: https://preset.io/blog/introducing-entity-centric-data-modeling-for-analytics/
Loved “old enough to have used Crystal Reports when it was a hot product” that’s me! 👍🏽
I’ve noticed, no doubt because of this discussion over the last week, that a lot of vendors are pushing the “modern data warehouse” and it feels just as vapid as the MDS. Like sure, it’s a lakehouse architecture and it’s more modern than your ERP, by decades, but that doesn’t actually solve any problems! It’s just fluff 🪶
This statement:
> The graveyard of dead or recycled terms in our field is massive, like “Big Data,” “Data Lake,” “Data Mining,” and so on.
makes no sense. Recycled? Maybe, I'd say improved. Graveyard? Absolutely not.