Installation Options for Process Integration and Orchestration Use Cases

[alex_bundschuh]

We recently have received many questions about the new Java-only installation options and whether the PI dual-stack is still supported or not. Rumors are spreading that the dual-stack won't be supported with the latest release. I have read this in many blogs even on SDN. This is actually not the case. In this blog I would like to take the opportunity to squash those rumors, and to help to shed light on the various installation options that SAP provides in the area of Process Integration and Orchestration.

First of all, SAP will continue to support the PI dual-stack with the upcoming releases. The PI dual-stack won't go away on the short term, at least it will be supported until extended maintenance of the NetWeaver release 7.3 which is until 2020. However, our main development effort is targeted at the Java-only installation options. New features, functionalities, and further enhancements will mainly go into the Java-only options. This does not mean that the dual-stack won't benefit at all from the latest development enhancements. Since the dual-stack comes with an Adapter Engine that runs on the Java plattform, some new developments will also be available here.

Secondly, we can tell that we are in a sort of migration phase currently from dual-stack to Java-only. And this brought up quite a lot of new installation options. I admit that we haven't been very fortunate in choosing proper names for the new installation options. So, all this ended up in the confusion we currently have. So, let me try to sort this out by providing you an overview of all relevant installation options that we have right now including main characteristic, supported releases, and use cases, see table and figure below. Note, that in figure below I have not included the PCK since it has been replaced by the AEX with 7.31.

Also, for a better understanding a bit of history. We actually started our journey towards a Java-only integration platform already with release PI 7.10. Here, we provided a possibility to bypass the Integration Engine in the PI ABAP stack during message processing, hence using local processing in the Adapter Engine. The intention was to reduce the hardware resources needed, the internal communication as well as the DB access points with the overall goal to increase the message processing performance in terms of message throughput and response time. This was possible for specific scenarios only, i.e., scenarios that did not need any capabilities that purely resided on the ABAP stack at this point of time such as various adapters and ccBPM. We liked to distinguish this new feature from a pure Adapter Engine, and hence from then on we called it Advanced Adapter Engine (AAE).

Within the last couple of years, we heavily invested in closing the gap between the AAE and PI. Whereas release 7.10 of the AAE only supported static routing, dynamic routing (i.e., content-based routing) as well as multi mapping support has been introduced with 7.11. In 7.30, a Java http adapter as well as a Java IDoc adapter have been implemented, just to name the most prominent features that were missing so far. The Java IDoc adapter covered the most important capablities that are supported by the ABAP IDoc adapter, full feature completion of the Java IDoc adapter compared to the ABAP IDoc adapter was finally introduced with 7.31.

Besides the AAE, from 7.30 on it was possible to install the Adapter Engine as standalone, the so called Advanced Adapter Engine Extended (AEX). From connectivity and messaging point of view, the AEX does not differ from an AAE. However, former comes with an own SLD, Integration Directory, ESR, Monitoring, UME, etc. So, it can operate on its own.

The only major feature missing on an AEX is supporting integration processes. On PI dual-stack, this is addressed by ccBPM which runs on top of the ABAP stack, and hence is not available on a Java-only server. So, with 7.31 we came up with the so called Process Orchestration (PO) which is a co-installation of the AEX and BPM/BRM usage types. In the long term, the PO will become our integrated stack offering where you can run any kind of scenarios, let it be stateless message processing, staful so called integration-centric processes or human-centric processes. With beginning of the 7.31 Ramp-up, the PO is however primarily targeted at exisiting BPM customers who like to benefit from the enhanced connectivity and mapping capabilities provided by the AEX. For integration-centric processes, we follow a different approach: we will deliver typical Enterprise Integration Patterns which are frequently used in real world scenarios. Those patterns will be shipped along with best practices, sample models, sizing numbers, etc., which will guide you through your implementation projects. It is planned to first ship the aggregator, content enricher, splitter, and resequencer patterns with 7.31 SP4 however this is subject to change. We do not like to throw this new option just over the fence, and then leave you alone, instead those patterns will help you to avoid common pitfalls during your implementation projects.

Having said this, I hope this led to a better understanding of the various installation options that we have right now, and also of where we came from and where we are heading to. The clear message should be that firstly we preserve your investment since we still support the various installation options side-by-side. Secondly, yes, we still support the PI dual-stack, but on a long term we are heading towards an integrated stack which runs on Java-only.

Table: Installation Options for Process Integration and Orchestration Use Cases.

Installation Options	Releases	Characteristics	Use Cases
Process Integration (PI)	all(1)	Dual stack installation Design and configuration: System Landscape Directory (SLD), Enterprise Service Repository & Registry (ESR), Integration Directory (ID) Runtimes: Integration Engine (IE), Adapter Engine (AE), Business Process Engine (BPE) Monitoring ABAP User Management Environment (UME)	Complete functional range Stateless message processing Stateful message orchestration As of release 7.31, improved federated PI capabilities via central ESR
Adapter Engine (AE)	≤ 7.0x	Connectivity: hosts most of the PI adapters Mapping runtime Cannot be operated on its own, i.e., design, configuration, monitoring , and UME via PI	Providing connectivity and mapping capabilities Deployment either centrally as part of PI server or as non-central option: scaling out, placing in DMZ, etc
Advanced Adapter Engine (AAE)(2)	≥ 7.1	Successor of AE Cannot be operated on its own, i.e., design, configuration, monitoring, and UME via PI / AEX / PO Enhanced to provide local message processing bypassing the IE leading to higher performance in terms of throughput and response time	Same like for AE Run synchronous scenarios in local processing mode at faster speed Move specific scenarios from PI to AAE for higher throughput
Advanced Adapter Engine Extended (AEX)(3)	≥ 7.3	Java-only standalone installation SLD, ESR, ID, and monitoring Runtime and connectivity capabilities of AAE No support for integration processes Java UME	Lightweight integration at lower costs Like PI, can be scaled out by adding further AAEs Federated PI, see above Replacing PCK as of release 7.31, see below
Process Orchestration (PO)(4)	≥ 7.31(5)	Java-only standalone installation Co-installation of AEX and Business Process Management (BPM) / Business Rules Management (BRM) SLD, ESR, ID, Message and Process Monitoring Java UME Design and Configuration as plug-in in Eclipse or NWDS	Reliable connectivity between AEX and BPM Calling PI mappings from within a BPM process As central instance only; can be scaled out by adding further AAEs Stateless and stateful message processing Federated PI, see above
Partner Connectivity Kit (PCK)(6)	≤ 7.3	Java-only standalone installation Mapping runtime Connectivity: RFC, file/ftp, JDBC, JMS, SOAP, Mail, XI; no IDoc Configuration, and monitoring Either inbound or outbound has to be connected to an Integration Server No support for integration processes	For small business partners or subsidiaries to integrate to PI of their business partners or head quarter Testing own developed modules or adapters
Adapter Engine Java SE (J2SE)	all(1)	Java-only standalone installation Based on J2SE, does not require a J2EE server Limited connectivity: file/ftp, JMS, JDBC, SOAP No Mapping Either inbound or outbound has to be connected to an Integration Server Remote configuration supported	Point of Sales (POS) scenarios

⁽¹⁾ Releases XI 3.0, PI 7.0x, PI 7.1x, PI 7.3x
⁽²⁾ Limited routing capabilities in 7.10, further enhancements in 7.11 and beyond
⁽³⁾ Minor capability gaps of IDoc adapter in 7.30 and other limitations, further enhancements in 7.31
⁽⁴⁾ Currently, released for existing BPM customers only covering human-centric processes with heterogenous integration needs
⁽⁵⁾ Ramp-up of release 7.31 has been started on 21st of November 2011
⁽⁶⁾ As of release 7.31, PCK has been replaced by AEX

Figure: Process Integration and Orchestration runtimes per releases.

I hope this was helpful information. In case of questions do not hesitate to comment on this blog or directly send me an email.