Types of integrations
The building of a successful integration solution begins with well-defined integration requirements. The requirements must explicitly define the integration type that satisfies the solution goals. As with most software-based solutions, the initial form of the requirements is the use case that represents the problem to be solved. The use case must be transformed into the solution requirements.
Integration use cases can be typed as data oriented, service oriented, or event oriented. Your integration type determines the methods you would use to accomplish the integration.
- With data-oriented integrations, large volumes of data are extracted from one or more sources, enriched or aggregated through transformation, and then stored into one or more destinations.
Data-oriented integrations are designed for querying, analysis, and reporting, rather than transaction processing. The data is usually historical and does not need updates. This type of integration requires ordered, predictable, and in many cases, compute-intensive processing. The processing is usually scheduled to start at predefined intervals.
Data warehousing is an example of this type of a data-oriented integration.
- In a service-oriented integration, some type of simple or complex orchestration is usually performed on multiple systems.
Data might be extracted from one system and used to make decisions about how to update other systems in the orchestration group. This is service-oriented integration because the orchestration platform provides an integration service to the collection of systems that participate in the integration. Any one of these systems can initiate an action by making a call to the orchestration platform.
An example of this type of integration is closed-loop compliance, in which a configuration item (CI) is found to be out of compliance by an audit job, which triggers an automated process to raise an incident. The incident can have aging states that can trigger escalation actions with respect to the CI.
- Event-oriented integrations are driven by events captured through the monitoring of external systems.
Captured events are usually standards-based, asynchronous-type events, such as SNMP traps or proprietary-generated events. A proprietary-generated event is an event whose format and communication protocols are explicitly defined by the host application. In this type of integration, processing a large number of events with high arrival rates is crucial. Events are usually designed to invoke a small number of discrete processing steps and can arrive in any order. After capturing an event, the integration solution might perform a simple action, such as sending a notification to a predefined list of targets, or a complex action, such as allocating or deallocating a service or resource. Usually, the goal of this type of integration is to ensure that all incoming events are captured and processed.
An example of this type of integration is the monitoring of network devices to capture and process fault and performance events.
Where to go from here
For more information about the integration capabilities of that you can leverage to fulfill solution integration requirements, see Integration mechanisms.