Integration is often the most complex part of a large ICT programme.
The defined, tested, monitored connections between applications, devices, databases, clouds and external agencies, so that independently built systems exchange data reliably and securely.
Systems may use different:
The integration architecture must allow these systems to exchange information reliably and securely.
Each layer has different failure behaviour: a device link drops intermittently, an agency interface changes without notice, a replication lag grows silently.
Integration may occur across:
The method follows the interface’s timing and reliability needs: synchronous calls for lookups, queues and streams where delivery must survive an outage.
Methods may include:
These twelve attributes form an interface specification. An interface with an undefined retry or reconciliation rule is an incident waiting for a trigger.
Each interface should define:
Testing covers abnormal conditions as deliberately as normal ones: bad, duplicated, late and missing data, and the loss of either end of the interface.
Integration testing should include:
Describe the requirement, the locations involved and where the project stands. We will respond with the technical questions that need answering first.