Agile Success Requires Enterprise Architecture

Hey! Wait a minute! Maybe enterprise architecture is not dead just yet! Check this out:

https://www.cio.com.au/article/626011/anz-seeks-partner-rethink-architecture-agile/

ANZ is one of the ‘big 4’ banks in Australia / APAC. Five million customers, AUD $250 billion assets. In the top 100 banks globally.

ANZ have adopted SAFe for all digital business technology innovation programs and projects. They love it. But agile does have a “learning curve”, doesn’t it?

The hyperlinked article above suggests ANZ understand that agile programs and projects require architecture support. And that requires an appropriate amount of architecture governance. And that requires an appropriate strategy for architecture knowledge management.

Gerard Florian, ANZ’s Group Executive Technology, asked:

“What is the best way to make sure we keep our architecture appropriately aligned but also flexible enough to evolve?”

Interesting that at some point in the past, ANZ implemented pricing functions into their CSC Hogan ledger system. (Probably because some architect thought it was a good idea…) So now ANZ have to de-couple pricing services before they can leverage blockchain. And that will take ten years. TEN. YEARS. Seemed like a good idea at the time. I’m sure other banks did the same thing.

We’re still heading to an architectural future where business services get even smaller, more atomic, more decoupled, more API-enabled, and more orchestrated.

And that spells “complexity”. And “complexity” is why they came up with … enterprise architecture.

The Object Oriented (OO) concepts of “service oriented architecture” (SOA) — which we all started working toward in the ’90s — have evolved — have moved up to enterprise scale — and are still relevant and aspirational today.

This pleases me greatly. I grew up with OO and SOA. And I’m not young.

SOA has been around longer than many vendor technologies and products that were considered “new and innovative” in the ’90s, but which are now considered “legacy”.

Microservices are just services. With the word “micro” tacked on the front. Because you want those suckers to be small. Small enough. And no smaller.

My point is:

As enterprises move to a decoupled, orchestrated-service-objects future, with agile and devops, we need enterprise architecture more than ever.

Ungoverned SOA is a known risk. It’s an existential threat.

But enterprise architecture — in order to be “agile” itself — requires discipline and science. A meta-model-based enterprise architecture repository is one key part of that science.

For enterprise architecture to competently support agile programs and projects, we have to “industrialise” EA so that we are able to give reliable, facts-and-evidence-based guidance to agile programs and projects.

Continually saying to business programs:

“We don’t have that knowledge. It will take us a few months to find out. We’ll get back to you.”

is no longer an option for EA teams. It never was.

The business is on the move. They’ll move ahead, with or without enterprise architecture.

Otherwise, enterprise architecture will be dead.