Cost Of Programming The Glue Code example essay topic

The cost of integration solutions is still dominated by the cost of programming the glue code that ties the integration business process (workflows and Web flows) to the underlying business applications. Unfortunately, the state of the art still remains a long way from nirvana -- when a business analyst will define only the high level business process, and the infrastructure will "magically" generate the necessary glue code. Historically, one of the biggest risks to integration projects has been their dependence on proprietary application programming interfaces (APIs) for programming this glue. With no standard APIs, there is no investment protection in integration programs or programmers. Nor are there the richer tools that follow from API standardization (such as those created for SQL and J 2 EE). Independent software vendors (ISVs) in particular are desperate to build integration solutions that can be deployed across multiple containers (e. g., be portable to both Web Logic and Web Sphere).

Of course, the Java and XML communities have been hard at work extending the Web application platform standards to protect investment in integration tasks. Some highlights worth a closer look: XML Query (W 3 C) Java Web Services (JWS) (JSR-181) Process Definition for Java (JSR-208) BPEL 4 WS (published by BEA, IBM, and Microsoft) Java Meta Data (JSR-175) XML Beans (open source from BEA) Portlets (JSR-168) Content Management Interface (JSR-170) Apache Struts and Java Server Faces (JSR-127) Web Services for Remote Portlets (WSRP; OASIS) Orchestration At the orchestration level, programmers include the declarative glue essential to integration: workflows, "Web flows,"work-lists" (collaborative document flow through an organization), and choreography -- abstracted message sequences that define how workflows can be composed within and even between companies. The most well understood technology at the orchestration level is Business Process Management (BPM). BPM refers to the graphical languages and tools for composing workflows -- computational sequences of tasks and exceptions. Most programmers are convinced that a variety of workflow representation languages will be required by the enterprise: Those targeted at the business analyst (e. g., Aris) Those that are seamlessly integrated with Java (Process Definition for Java / JP) Those that are programming language independent (BPEL 4 WS) For Java-based workflows, PDJ will remain the best fit.

BPEL 4 WS, however, provides a level of independence between the Java and. NET worlds, at the cost of introducing a new XML programming language. User should expect PDJ and BPEL to converge going forward -- PDJ will become the Java realization of a BPEL workflow Integration Brokering and Management While integration solutions still require coding, the goal remains to minimize that coding. Security is similar in this regard: historically, authorization was hand-coded into the application. Now the leading Web platforms define security policies administratively through business-oriented rules (that can be applied to a particular endpoint or across an entire application). The same shift is already underway for command and control -- the decision about how to route a particular request (say a trade) could depend upon the client (quality of service obligation), the content (dollar value), the receiver (which transform or adapter to apply), or the state of the infrastructure (availability, load).

If such logic is programmed into the application, then it cannot be changed without modifying and redeploying the business logic. By capturing such metadata administratively, change can be accommodated far more easily and economically. The integration platform's management environment must provide a comprehensive operational view into the inter workings of the integration broker and its connections to various applications. Looking forward, integration environments will have to become more self-optimizing and self-healing -- raising alerts to the human administrator only when the available resources are insufficient to meet the committed quality of service.