I now know of two global insurance companies who started their digital initiatives by trying to define a global, canonical data model, so I guess my two-year-old blog post is still relevant: innoq.com/en/blog/thoughts-o… TL;DR: You don’t want to do that.

Why You Should Avoid a Canonical Data Model

As an enterprise architect, you might be tempted to strive for a canonical data model for your systems’ interfaces. That’s not a good idea.

innoq.com

Nov 8, 2017 · 1:51 PM UTC

16
145
21
236
Load newest
Replying to @stilkov
bigger orgs tend to like the idea of a perfect #EIM. my answer to them these days: your #API landscape *is* your #EIM. it’s as good as it gets.
1
Replying to @stilkov @martinfowler
Perfect model: class Model { String key; String value; }
1
9
Replying to @stilkov
+1 for “The amazing thing is that organizations are excellent in generating a huge amount of work based on bad assumptions.”
2
Replying to @stilkov
Global is not good enough, you need universal 😀
1
1
Replying to @stilkov
There aren't many bigger enterprise integration exercises than niem.gov which has been a spectacular success. Your "how-to" points point to the kind of approach that has worked. CDMs must be faceted and advisory, not enforced on every subsystem architecture.
1
Replying to @stilkov
Ha, tried going down that rabbit hole in 2008.. good post 👌
1
Replying to @stilkov @vmx
Agree 100%. Saw this failing a few times already. But the CDM idea is simply too appealing.
1
Replying to @stilkov @martinfowler
But, but, but... How is their ESB supposed to work without a canonical data model?
3
Replying to @stilkov
youtube.com/4U0AqxlxfBs #notbraveenoughtotagpeople
This tweet is unavailable
I agree, but you could view this as an instance of true true scotsman fallacy …