The evolution of integration
Modern business processes like order processing or warehouse planning often rely on multiple, disparate software applications working together as a whole. Integration is the art of making that happen. In the past, this was handled by middleware running on-premise.
But as companies move towards a best-of-breed approach with numerous highly specialized applications and modules, the need for a more performant and flexible integration solution emerged. Hybrid integration platforms were born. Traditionally, these platforms combined on-premise and cloud integration. The modern versions also include an integration layer and a data layer.
- The integration layer is where data from various applications is collected or retrieved.
- The data layer is where your company data model is centralized.
Avoiding data spaghetti
Why a data layer? In recent years, data emerged as a key driver for business success. As companies started using more business applications, however, data became increasingly scattered across the organization. Customer data, for example, is no longer confined to a single tool, but can be found in your CRM, ERP, and a plethora of other applications. Getting the right insights for a specific use case now requires seamless sharing and synchronizing of specific data pieces across multiple components.
As a result, the role of the ‘integration specialist’ has expanded from being a mere interface builder to taking a holistic perspective. For every integration, they now need to consider which data is needed where, what its purpose is, which application owns it, etc. And while this would technically be possible to manage with point-to-point integration, data is so deeply interwoven in the modern enterprise that the result would be an unmanageable ‘spaghetti’ – hence the added ‘data layer’ in modern hybrid integration platforms.
4 integration styles
Data is, of course, not the only thing that’s important in integration. Broadly speaking, there are four different styles of integration, each with their own requirements, characteristics, and matching tools.
- Data integration (see above) focuses on delivering consolidated, consistent, and high-quality data between different applications. For example, ensuring customer master data is consistent between ERP, CRM, and marketing applications.
- Process integration focuses on collaboration of the various applications involved in a specific business process. Your production system, for example, needs information from many different applications to function, e.g., what should be produced, by when, etc. In process integration, message integrity is key.
- User integration focuses on catering to the data needs of business users, end users, and customers.
- Things integration focuses on making IoT devices and the massive amount of data they generate work well with the rest of your IT landscape.
These integration styles also represent the four main areas of volatility in integration. Each of these aspects is subject to change and needs to be encapsulated to avoid the integration setup to fall apart. Knowing the desired outcome of the integration is therefore essential.
APIs and events
Luckily, modern hybrid integration platforms make it relatively easy to deal with volatilities. There are two main approaches, which are often used in combination.
- API-driven: Most modern applications come with a fixed set of APIs to trigger specific actions in other applications. These can be freely used, and don’t require any custom development.
- Event-based integration: Instead of sending out a complete message immediately, the application sends out change notifications. If other applications want to collect this changed data, they can use the first application’s API.
As mentioned before, companies have moved on from using all-in-one business solutions to picking specific applications for specific business process or needs. Interestingly, the same thing is also happening in integration platforms themselves.
In the SAP BTP Integration Suite, for example, every step in the integration process – receive, persist, route, transform and distribute – can be handled by different components. This also means that it might not make sense to acquire the entire integration suite: just a few specific components might suffice, depending on the maturity of the customer’s integration landscape and requirements.
Pillars of the modern enterprise
Integration platforms play a vital role in the functioning of the modern enterprise. By acting as a central repository and replacing classic point-to-point connections with a loosely coupled integration platform, relying on API’s and event-driven architecture, they ensure reusability and continuity. By isolating processes, data and users, these platforms allow companies to ‘keep the core clean’, no matter how many individual business applications they introduce or changes they make. Moreover, knowledge of specific integrations is no longer only stored in the head of an individual expert but can easily be transferred to others.
Every integration process requires careful consideration. What are the volatility risks? Which building blocks should you use? How can you make it future proof? Just like enterprise architects, modern integration experts or consultants need to have a holistic view of your landscape. Ideally, they’re involved in the project as soon as possible to mitigate any risks and create more flexibility further down the road.
Get in touch with our experts to discuss the possibilities for your company.