SAP high availability

You can configure IBM® Integration Bus to withstand software or hardware failures when working with SAP, so that IBM Integration Bus is available for as much of the time as possible.

In previous versions of IBM Integration Bus, the tRFC protocol between SAP and IBM Integration Bus (acting as the RFC server) ensures that IDocs and tRFC BAPI calls are delivered exactly once. This behavior is possible because each delivery has an associated transaction ID (TID). IBM Integration Bus monitors the progress of a delivery until SAP confirms a successful delivery. If the connection is lost or IBM Integration Bus fails before that confirmation is issued, SAP attempts to redeliver the message. By keeping a persistent record (in the TID store or transaction log), the integration node can ensure integrity and avoid a duplicate delivery.
This diagram shows that a translation log keeps a persistent record.
When two .inadapter components, with the same RFC program ID, are deployed to two integration nodes, two connections to the same RFC server are visible to SAP. If the connection is lost to one of the integration nodes, SAP might attempt to redeliver to the other integration node. The integration nodes have separate TID stores, therefore, the second integration node accepts the redelivery, even though the first integration node might have processed some (or all) of the IDocs in the packet.
This diagram shows separate TID stores for two .inadapter components.
In IBM Integration Bus, you can now move the TID store to a remote queue manager that can be shared between two integration nodes. To avoid a single point of failure, make this third queue manager a WebSphere® MQ high availability, multi-instance queue manager. For instructions, see Setting up SAP for high availability.
This diagram shows how a third queue manager is used to avoid a single point of failure.