What does a coreless business look like?
A centralized business has its IT architecture – storage, compute, programs, databases, etc. – organized and managed centrally. Each part of the business draws on centralized or siloed platforms for the software or data it needs.
This architecture will often be a jumble of connected systems, built around monoliths – major business applications that combine multiple joined-up systems into a single unit. An example would be an Enterprise Resource Planning (ERP) system that integrates inventory management, production scheduling, and order processing into a piece of software that runs on a single server. Any update, no matter how small, requires redeploying the entire application, followed by extensive testing.
A coreless architecture, on the other hand, has IT designed around its business functions – e.g., manufacturing, supply chain, production, marketing – or by product line and domain (i.e., a grouping of IT devices associated with a particular business function and controlled by a single set of rules).
Each business functions around a domain, each of which includes a series of microservices: e.g., one for inventory management, one for production scheduling, one for order processing and so on. The business function combines these services as needed to achieve its goals. However, rather than being connected, each microservice operates independently and communicates with one another using APIs, leveraging a ‘service mesh’, and ‘data mesh’ – infrastructure layers that allow disparate apps to communicate across siloed IT systems.
Therefore, each microservice can be updated or scaled independently, as needed, without impacting the rest of the system. New services can be easily added to the IT setup.
So, say a manufacturing team wants to launch a tool to analyze social media and spot reports of product defects, which can then immediately be addressed in manufacturing. In a coreless setup, they simply develop a microservices-based product and launch it within their own IT architecture, without lengthy testing or risk to other systems.
By decoupling the business functions from underlying IT systems, coreless architectures enable teams to make decisions without being constrained by IT limitations. In real terms, that means digital products and services that previously took over a year to release could now be launched in weeks.
Decentralizing: Abstraction vs Coreless
A ‘middle ground’ is often proposed as a solution to the ‘Centralized IT’ problem. This involves layering a data mesh and service mesh on top of the old architecture, to abstract away the complexity by creating virtual instances of different applications that can be controlled independently. This is an improvement, but not a perfect solution, since the old underlying architecture is still there.
What we propose in this paper is something radically different: pulling everything apart and reassembling it in a way that is genuinely decoupled.
AWS: Lessons from the best
AWS is the most profitable part of Amazon, one of the most successful companies in the world's history. Its coreless approach makes this possible, and AWS owes its success to continuous innovation.
AWS began small with three product lines: storage (S3), messaging (SQS) and computation (EC2). These were designed to the same standards – so they can interoperate, but are standalone services. Amazon has since built a range of higher-level cloud products, from databases to machine learning.
For customers, these products can be combined to create a bespoke cloud service. Internally, they are distinct product lines with their own IT setups, and their own teams, who have the autonomy to make changes and upgrades without affecting any other products. This makes it easy for Amazon to continually improve its legacy products, launch new ones, and integrate acquisitions.
Unlike many large enterprises, it does not reach a point where it gets stuck. It has market share, legacy infrastructure, and the ability to constantly innovate. That makes it very hard to dislodge.
