As integrations and APIs scale, the traditional central integration delivery team quickly becomes the central bottleneck team. Overwhelmed by demand, it becomes overworked while the pace and scale of delivery slows for IT organizations dependent on it. Tech debt also increases: To work around the bottleneck, development teams may leverage their own integration products or build on whatever product’s support team has the shortest backlog rather than the product that is the best architectural fit. What you need is a federated delivery of integration. Roche sets the example.

Roche’s Digital Integrations Team Provides The Model

To learn more about Roche’s delivery model, I met with its digital integrations team, which is responsible for integration infrastructure. I initially planned on just exploring their automation via platform engineering. Yet the discussion quickly showed that the big picture is federated delivery. Platform engineering was just one piece of the federated delivery puzzle. Indeed, platform engineering alone is not enough. You must situate platform engineering within the right broader operating model.

Roche has platform engineering teams for each of its five integration stacks (API management, event broker, iPaaS, and more). But central to Roche’s model is integration services and enablement (ISE), a team within digital integrations that serves as a center for enablement. It oversees the entire integration portfolio rather than siloed enablement teams separately for APIs, events, iPaaS, etc. ISE approaches development teams like the customer success division of a vendor. This includes tracking customer satisfaction metrics and establishing an “integration champion.” It aligns each integration champion to development teams. They then build expertise within those development teams so the teams can deliver integrations on their own. The champions also bring feedback to the five platform teams to shape the direction of the various integration tools. This structure is the foundation of a federated delivery model: empowering development teams to build and govern their own integrations, guided by the expertise of the ISE and supported by the automation of the platform teams.

Federated Delivery Produces Results

Roche has seen significantly faster speed to market and saved hundreds of hours of labor from platform automation. Its federated delivery has enabled massive scale of delivery: It averaged 10 API deployments per day with 1,300 API deployments to production in the last six months of 2023. That’s an impossible feat with centralized API delivery — and that’s just looking at its API stack! ISE has also facilitated tech debt reduction. It monitors unused assets and steers people to the right integration tool rather than the familiar (and legacy) tool.

While Roche’s model isn’t a perfect fit for everyone (no model is), it sets the example for those who need a rapid pace of delivery at massive global scale. Organizations whose needs are slower or at smaller scale can still learn from Roche by identifying which parts apply to them and which don’t. Even Roche recognizes this: Some of Roche’s business units are too small and lack the resources to take up federated responsibilities. These units continue in a more centralized model.

Roche’s Integration Approach: Ahead Of The Curve

This wasn’t my first time meeting with Roche. A couple years ago, I interviewed the same team on their forward-thinking approach to converge event-driven architecture and REST APIs. I have cited that report many times to my clients, and I expect to do the same with this new one. There’s much more to Roche’s federated delivery model than what can fit in this blog. Forrester clients can read more of what Roche is doing and best practices to apply in this latest case study. Both show that Roche is ahead of its peers in its approach to integration.