Integration

Traditional message or file-based event-based integration where a system occasionally generates a transmission to one or more recipients.

A message = an information object such as an order or a list of several such objects (e.g. an exported batch). Characteristic of this type of system integration is that complete messages are sent, either directly between systems or, ideally, via a central MOM (Message Oriented Middleware) that provides routing (the right message to the right recipient), aggregation and transformation (mapping).

For more dynamic information flows, event-driven solutions can be used. In such solutions, systems send "signals" to an event hub with "events", such as Order Created and Shipping Complete. Often this is combined with different forms of sourcing patterns where messages sent are very small and only describe what happened, where and when. The sender has no control or knowledge of the receiving system. These connect as subscribers to these events and can then perform their own logic when certain important business events occur. They can also retrieve information from the source to find out, for example, the content of the order, without having to send it in a large message through a message bus.

This provides increased flexibility and very loosely coupled solutions where producers of events, as well as consumers, can be replaced or added over time without costly changes. It also facilitates requirements gathering if you can analyze your business processes from an event perspective, which usually leads to interesting business insights.

A major advantage of serverless and loose connections is that you can isolate "steps" in processes to their own logic. These can be maintained separately and have their own life cycle, which further decouples things in the platform and reduces change impact. Then you can work with different tools to "orchestrate" flows by connecting these different steps with each other, to different types of information processes. For example, you can receive an image, send it to another piece of logic that creates a thumbnail, then to another that uploads the image and a text on LinkedIn. If you need to change the size of thumbnails, you can then change the isolated function and without impact continue to run your flow as usual in the meantime (minimal change impact). For this to be possible, different types of orchestration tools are needed. For example, you may need to restart a sub-step. Another step might be asynchronous (we don't wait for the result but expect it to come later). Yet another might need input in the form of crypto keys, which we want to retrieve from a key generator. All these steps can be combined with such logic into "flows" that we can then deploy, configure and monitor as a unit. We have extensive experience with such tools and solutions and can also build our own management if necessary. We also often build the various sub-steps that are then reused in processes. It is very effective for businesses to break down their various process steps in this way and it allows you to "break down" larger problems into smaller, reusable sub-solutions. Orchestrations can also be advantageously linked to MDM to internally work with "master objects" (internal representation) and then in the final step transform to different designs depending on the consumer of the process output.

All other integration types can be fully or partly API-based where an endpoint is an API. However, here we mean a pure API management for integration and should be managed by an API management service but has the characteristic of dividing into API layers depending on the rate of change and area of use.

Ex: Business-critical system APIs should have a low change rate, but if an agile organization requires a mobile API for a quick POC (Proof Of Concept), a custom API for this is implemented on top of the business-critical API. More and more systems today have their own APIs (often web-based) that can be used for information retrieval and thus system integration. However, these may be based on standards that are considered "unsafe" or, for example, lack support for REST-based communication. Then we can help centralize, secure and package APIs from different systems through an API Management platform for central monitoring. This can then be used to create access to APIs without giving external actors direct access to underlying business systems. A very powerful and useful solution that creates opportunities to share key information with others in a secure and efficient way.

Integration of "data in rest", i.e. data that does not need, or should, be transmitted in real time.

Often large batches of data that are scheduled with periodic transmission.

E.g. common for transferring data to DataWareHouse or Data Lake. Here we often work with different types of information streams. From the "hot" stream where real-time information flows to the "cold" stream where we "wash" and store information for a long time. By building smart solutions, you can leave traditional OLAP and move towards analyzing the data that flows in the organization directly and let information that is already integrated drive data science and automated business decisions. We provide the necessary expertise to build and prepare solutions for modern data science and analysis of streaming data in real time, as well as support in leaving the limitations of traditional databases.

Master Data Management - centralized function and capability to ensure a single trusted source for each information entity. To avoid duplication and data inconsistency between systems and databases.

Integration between device and internal IT. Management of telemetry. Small messages with measurement data sent at high frequency and consisting of many data points. Requires own solution architecture and software for good performance and high availability but benefits from sharing know-how and services with other integration types. We provide the expertise to build modern solutions that connect physical devices with business systems and analytics tools over a variety of protocols such as MQTT, AMQP, LoraWAN and Modbus.

If needed, we can develop more complete enterprise middleware that includes specialized functionality, logic and information transformation. These are complete, tailor-made platforms, with support for serverless functionality, orchestration of information flows, API management, virtual networks for connections between e.g. cloud and on-prem, central security, encryption and key management, and analysis tools. We have delivered solutions in this area that are at the absolute forefront of technology, vendor agnostic and highly competent. Often this work is combined with supporting strategies, business objectives and architecture patterns in order to build a holistic support for the business.