Product Review

Digg This!

From a broad perspective, the purpose of any business software application falls into one of two categories: operational or strategic. Operational applications provide users with the capability to carry out business processes (such as processing a customer order from order receipt to product dispatch), and present users with the ability to manage the business data accompanying these business processes (such as customer information, order details and delivery status). Strategic applications (such as data warehousing and business intelligence) provide users with the ability to understand and analyze data in order to learn from it. In order to succeed in these purposes, it is essential that the application have access to all the data it requires. Historically, this data was typically owned by the application, and located within the "application silo." Over the years, however, there has been a growing need to reach out and acquire data that lies outside the control of the application, and today, with mounting volumes of diverse and disparate data, data integration is a keenly felt need and a widely accepted mantra. Liquid Data 8.1 for WebLogic from BEA Systems is a data integration solution that is particularly relevant to the Web service-centric world of today, and a solid contender in the J2EE data integration space.

Liquid Data enables users to create a virtual data source that consists of data attributes pulled in from numerous data sources using XML, and then publish methods to manage this virtual data by means of a Web service, served also in the form of XML.

Liquid Data Architecture
Liquid Data sits atop of and is tightly coupled with the BEA WebLogic Application Server. Figure 1 shows the Liquid Data architecture. The main components are:

• Data View Builder: A GUI-based tool for designing, generating, testing, and deploying Liquid Data queries. The Data View Builder provides drag-and-drop functionality to select and bring together data elements from a variety of data sources, including relational databases, XML files, csv files, Web services, and other applications (through WebLogic adapters). Data sources first need to be registered using the WebLogic Liquid Data Administration Console, which includes the functionality (i.e., additional tabs) for Liquid Data. The selected data elements combine to form a schema that provides a virtual integrated data source. Data View Builder generates stored queries in XQuery to manage this virtual data, as well as data views - i.e., queries that can be used by other queries and provide an additional level of abstraction - which can be then stored in the Liquid Data repository. Client applications can now use the stored queries as well as execute ad hoc queries, without regard to the underlying (and potentially distributed and heterogeneous) data sources. All in all, the Data View Builder provides the essence of data integration (see Figure 2).
   

• Liquid Data Repository: Stores the data views, queries, source and target schemas, Web service descriptions, and other such artifacts.
• Liquid Data Server: A J2EE application running within WebLogic. The server receives queries (in XQuery) from client applications and executes these queries against the base data sources, using the metadata stored in the Liquid Data Repository and an optimized query plan generated by a distributed query processor within the Liquid Data server.
• Liquid Data Query API and Liquid Data Control for WebLogic Workshop: An application programming interface for client Java applications needing to access data published by Liquid Data. Enterprise JavaBean (EJB) and JavaServer Pages (JSP) clients can invoke queries using the Liquid Data query EJBs and the Liquid Data JSP tag library, respectively. In addition, applications developed in WebLogic Workshop can directly use the Liquid Data control extension to the Workshop IDE.

The User Perspective
All this sounds good, but what does it present from a user perspective? The answer to this question depends on the user. Data architects now have a tool that provides a ready means to integrate disparate data using a drag-and-drop paradigm and without needing to acquire advanced skills in J2EE and XML technologies (although my advice to data architects is to get started on acquiring some measure of these skills anyway!). Application architects will be encouraged by the support for a Web services and standards-driven approach taken by the product team, while systems administrators will benefit from the tight integation between Liquid Data and WebLogic, thus enabling seamless administration. Developers will find the Liquid Data control in the WebLogic Workshop IDE convenient, and they will particularly appreciate the sophisticated code-generation capabilities, which can significantly reduce coding for such related tasks as Web services and data access and excavation. Finally, performance engineers may want to avail of the query caching feature, which provides potential performance benefits.

The benefits of an information integration tool such as Liquid Data are not restricted to the technical team, however - in fact, the ultimate benefit is to the business community. The ability to integrate conventional data sources such as relational databases and mainframe applications, together with more contemporary (may I say post-modern, as in art?) sources such as XML and Web services, will help enterprises protect existing investment as well as leverage it to gain greater information awareness and competitive advantage.

While there are numerous benefits in using Liquid Data, there are other considerations and concerns to be aware of. For example, tight integration of Liquid Data with the BEA WebLogic application server will restrict using Liquid Data with other application servers. In addition, the Liquid Data release notes mention limitations in the ability of Liquid Data to support certain vendor-specific relational database features this may be an inhibiting factor for widespread deployment. Data mapping is a complex endeavor and requires correspondingly complex capabilities. Data mapping in the Data View Builder also accommodates data transformation, typically required to map different definitions of the same data attribute across disparate systems/sources. Finally, it appears that Liquid Data is designed to render data in one direction only - from data source to consumer, and cannot be used for data updates or data maintenance. (Editor's note: While Liquid Data alone does not support data updates, they are possible within the same Web application or Web service using other components of the BEA Platform [e.g., Workshop controls and/or WebLogic Integration processes for complex cross-source updates].) This has both advantages and shortcomings, and should be considered in a case-specific manner.

Conclusion
On balance, BEA Liquid Data is clearly a square peg in a square hole. The product is well designed, fulfills its stated purpose well, and satisfies an immediate business need. For J2EE shops in general and BEA WebLogic shops in particular, Liquid Data is a must-see, if not a must-have.

Acknowledgements:
My thanks to Anand Barhate (anand_barhate@hotmail.com) for his help in deploying and testing Liquid Data for this review.

BEA Systems, Inc.
2315 North First Street
San Jose, CA 95131
800 817.4BEA (U.S. toll free)
408 570.8000

Web:
www.bea.com

E-mail:
sales@bea.com

Requirements:
BEA WebLogic Platform 8.1