Problems
The characteristics of the above-mentioned use cases are as follows:

• Requires integration of a user’s personal content with enterprise systems – emails with an enterprise sales system
• Requires personalization for the individual user – integrate with Google Maps or Yahoo Maps
• Requires simple and non-programming ways to integrate – enterprise users are not JavaScript or AJAX geeks
• Most of the integration endpoints are present or coming up in an enterprise environment quickly
• Most of these integrations are “situational” – they don’t have a long-term value for the user or enterprise
• Decentralization of the IT environment – enterprise users are empowered to create and maintain their services

Enterprises are looking to decentralize their IT environment and are trying more and more self-service applications for their employees and customers. Most of the time enterprise-wide, self-service applications (e.g., expense reporting, sales dashboard, daily business intelligence including enterprise portal) are complicated and not personalized enough. There seems to be a rising enterprise appetite for situational and personalized applications reusing existing enterprise services and information assets. Until recently, the focus has always been in the service provider side – SOA, EAI, etc. But now, enterprises are looking for more personalized services for a more productive workforce and demanding customers.

Even though SOA enables business applications to interoperate with each other, the consumption of those combined applications is still not high by enterprise users for their personal decision-making and workflow automation. Rich Internet applications (RIAs) built with Web 2.0 technologies are only one set of interfaces for enterprise applications; personal desktop applications like email, calendars, or spreadsheets need to be included in the list of SOA applications for consumers to realize the higher ROI of enterprise SOA investment.

Mashup technology is quite popular for combining different Web-based applications available over the Internet, but their usage is still confined to experts of JavaScript and Web 2.0 and is mostly for non-enterprise, Web applications.

SOA-Based Enterprise Mashup Platform
Mashups and SOA can be brought together to solve the enterprise composite application puzzle. To make it successful, we need to have a distributed, widely accessible service or information mediation framework that will support declarative ways of defining service composition for non-programming enterprise users and will empower every enterprise user to create their own composition logic with enterprise and/or personal information sources.

SOA is making enterprise mashup possible by enabling all enterprise information sources to be accessible to every user. SOA-enabled applications and information sources can provide a ubiquitous computing environment. Soon all enterprise IT environments will be full of services exposed through standard SOA interfaces. If the enterprise users can combine these services easily, then only enterprises will be able to reap the benefits of SOA investment. Mashups are becoming an established technology for achieving the same, at least with the Web applications.

SOBA – coined by Gartner –are considered to enable enterprises to dynamically integrate applications according to the business needs and situations in real time. Enterprise Mashups are an extension of the same vision where integration takes place not just within the enterprise but also beyond – partner services, external Web services and JavaScript/RSS/Atom.

Our proposed platform will allow users to combine the different enterprise SOBA applications and personal information sources and services to create new personalized “situational” composite applications. This service mediation framework will be rule and model based and will stand for mediation systems for future enterprise information systems. User will be creating these new composite services with the help of a browser-based rich client simply by specifying the services to be connected and optional aggregation, transformation, and filtering rules. This platform will be a distributed, hosted framework and will execute personalized, user-centric service composition rules. There will be a standard service registry (e.g., UDDI) attached with this platform, which will contain all available enterprise services. The user will be able to model his/her service composition logic in a declarative manner by selecting personalized or enterprise information sources for the combination. These newly created composite applications can be combined again to create a higher-level composite application.

Enterprise mashups can’t be just a mere composition of services, but they have to come with reliability, security, and effective usage of enterprise resources. All the attributes of an enterprise system such as security and location transparency will be completely encapsulated in the framework. As enterprise mashup solves a particular business problem – tight coupling with a particular service needs to be avoided as the service may fail capriciously. Our platform will provide dynamic service lookup and dynamic composition to provide loose coupling, high availability, and location transparency. Figure 1 shows the high-level architecture of the framework.

Rule-Based Declarative Enterprise Mashups
Rule-based or declarative mashups for external Web applications are useful but for enterprise users (who hate any form of coding) this is a must-have, to be acceptable. Declarative service mediation or composition has been the buzzword for the SOA products for sometime. Enterprise Service Bus, BPEL, etc., are examples of declarative rule-based SOA. Rule-based systems are easier to develop and maintain. The same principle can also be applied to enterprise mashups as well. Our proposed service mashup framework will standardize simple rules to provide the mixing and matching of those available enterprise services for every user. Users will be able to declare rules for the aggregation, transformation, and filtering of the contents of different information sources. For example, Intel’s MashMaker has been able to create a rule-based mashup platform with interactive UI.

Personalized Enterprise Mashups
Enterprise users need to be allowed to combine information from enterprise search engines, Web services, messaging systems, business intelligence engines, and data integration solutions, along with information from their partners and external sources. SOA is meant to provide the ubiquitous presence of services from every possible information source. But what is lacking is user empowerment to create applications that suit their needs. Personalized enterprise mashups are just-in-time enterprise information integration but per user, by the user, and meaningful to only the user for that “situation.”

This proposed framework will allow user-specific service composition rules and integration with their personal information sources and applications. That way, users will be able to create their own sales dashboard from enterprise sales applications that are also linked with their own personal sales spreadsheet and Google Maps for regional information. Sharing personalized mashup plans with co-workers or tagging useful enterprise services will be the collaborative features of the platform.

Future Work
In this article, we described how enterprise information mashups could be materialized with a rule-based service mediator framework. One interesting extension will be a grid-based mashup platform to handle the sudden increase of developer cum users.

Conclusion
A SOA-based distributed and hosted service mediation framework will help enterprise users build up information mashups for their own situational applications and will enable lightweight information and service integration. The framework is not a replacement of traditional enterprise information integration, but an extension that makes integration personal to users. This framework will be a simple rule-based creation of those personalized enterprise mashups that will help the non-programming business user crowd to use the enterprise information sources more effectively and in a timely manner.

Reference
Intel MashMaker: http://mashmaker.intel.com