Diasoft explains why it built the Digital Q.Integration platform instead of an ESB

Viktor Ovchinnikov, head of development for the Digital Q.Integration integration platform at Diasoft, explained why his team abandoned the classic ESB (Enterprise Service Bus) architecture and built its own approach to integrating enterprise systems.

Why the classic ESB didn't fit

According to Ovchinnikov, a classic ESB is built around a monolithic core that scales only vertically. You can boost performance only with more powerful hardware, not by adding new nodes. If that core fails, all integration stops: none of the connected systems can exchange data with any other.

But the most painful problem for operations, in the author's view, doesn't show up right away. It builds up gradually. Over time, the bus becomes the place where all of a company's application business logic ends up. After a few years, the ESB knows more about the company's business than any of the systems connected to it, and untangling that logic becomes a hard problem in its own right.

The approach Diasoft's team chose

The team set itself a different principle: take the best of point-to-point architecture and the best of the classic ESB. On one side, independent services with no centralized business logic, as in direct point-to-point integrations. On the other, shared transport, shared monitoring, and shared tooling, as in a traditional bus. The team named the concept "smart services and reliable channels."

• A classic ESB scales only vertically and is a single point of failure
• Over time, an ESB accumulates all of a company's application business logic
• The Digital Q.Integration platform is built on the "smart services and reliable channels" approach
• Each integrated system gets its own dedicated connector microservice
• Adapters communicate through a shared transport layer responsible for delivery and reliability

How the platform's architecture works

For each integrated system, the platform creates an independent connector microservice. It contains exactly the logic needed to work with that specific system and nothing more. The connector knows nothing about other systems' business processes and doesn't turn into a centralized repository of knowledge about the company.

Adapters talk to each other through a separate transport layer that handles message delivery and the reliability of data exchange. This split resolves the core contradiction of the classic ESB: the bus used to handle both business logic and transport, mixing two different kinds of responsibility in a single monolithic core. Diasoft's approach separates them. Business logic lives in independent microservices, while a shared layer takes care of delivery and reliability.

How this differs from plain point-to-point

Direct point-to-point integration solves the monolithic-core problem: it is distributed by design, and the failure of one link doesn't take down the whole network. But point-to-point has a downside. Without a shared transport and monitoring layer, the number of direct connections between systems quickly turns into an unmanageable tangle, and tracking which system exchanges data with which becomes as hard as untangling the logic of a monolithic ESB. Diasoft's approach keeps the distributed nature of point-to-point but adds a unified transport layer and shared monitoring, both of which are missing from a pure point-to-point architecture.

What this means

Diasoft's approach is an attempt to fix the main pain point of classic ESBs: the bus turning into an opaque monolith that, over time, knows more about the business than the integrated systems themselves. For companies building AI projects and other digital services on top of many enterprise systems, independent connector microservices reduce the risk that the integration layer itself becomes a bottleneck and a source of opaque, hard-to-maintain business logic.