Skip to content

Is the modern data stack disappearing?

Today we answer the most important question. Is the modern data stack coming to an end?

Christophe Blefari
Christophe Blefari
2 min read

No.


This question generated a lot of content last week, and a lot of words were written. I wanted to keep my answer short so as not to burden you with a few thousand more words to read.

  • Modern data stack has been coined by US companies and VCs—mainly Fivetran / dbt Labs—as a word to quickly emphasis a way to build data stack in the cloud related to ELT. It was a well-suited marketing term, let's be honest.
  • The time came and everyone took their place at the table to eat a slice of cake.
  • A lot of people have issues with the modern word. Probably because it's not an explicit semantic, definition is relating to the present or recent times as opposed to the remote past. In this definition there are 2 issues.
    • is relating to the present — not all companies are in the same present
    • as opposed — the term creates an opposition between 2 worlds, creating something we always like in tech: a debate between 2 kind of technologies.
  • Actually modern creates some kind of exclusion between new technologies and old technologies. It was useful as first for Fivetran or dbt Labs to be disruptive, but now that everyone is using the MDS is it still a good idea to create this competition? Especially if you want to enter the Fortune 500 where they actually use old tech?
  • And, we should stop being cynical, who in the hell—in my readers at least—wants to work with SAP, IBM or mainframes in 2024? Because they still exists, numbers show that still 50% of companies are on-premise, when it comes to publicly listed companies or government stuff it's probably way higher.
  • For these organisations the ideal of a modern data stack still resonate. Employees are stuck in hell regarding data tooling. Data projects are still failing to go in production.
  • Personally, down there, my vision of the modern data stack has always changed over the years. As always, I don't blindly apply the principles by the book. The idea of dedicated storage where all the data lies with SQL transformations on top and top-notch CI/CD processes with everything-as-code and a galaxy of convenient tools around to be observable sums up what's modern about our data ecosystem.
  • That's why I think modern data stack vision isn't going anywhere.

In my four years of freelancing, I've always said I build data platforms or data stacks, because who am I to judge whether I'm modern?


As a reference read my online friends views:

  • Is the "Modern Data Stack" still a useful idea? — Tristan Handy, dbt Labs CEO. He mainly coined the term and whistles the end of the playtime. MDS was previously useful to align practices but now he thinks we should move on to analytics stack. And AI is around the corner to take all the lights while we actually do stuff at the bottom of the pyramid.
  • The problem was the product — Benn Stancil, Mode CTO, scroll to mid-article.
  • Everything Ends - My journey with the modern data stack — Joe Reis, author of Fundamentals of Data Engineering. Joe depicts his own journey and views and why it became a mess with too many companies on the radar. Creating finally the most fragmented platforms with no coherence at all, negating all the good MDS aspects.
Modern data stackdbt

Data Explorer

The hub to explore Data News links

Search and bookmark more than 2500 links

Explore

Christophe Blefari

Staff Data Engineer. I like 🚲, 🪴 and 🎮. I can do everything with data, just ask.

Comments


Related Posts

Members Public

dbt multi-project collaboration

Use cross-project references without dbt Cloud. This article showcases what you can do to activate dbt multi-project collaboration.

Members Public

How to get started with dbt

What's a dbt model, a source and a macro? Learn how to get started with dbt concepts.