Arch2Arch Podcast 6: What does Web 2.0 mean for Enterprise IT?by Ajay Gandhi To explore what Web 2.0 really means for enterprise IT, Ajay Gandhi, director of enterprise social computing, asked the architects behind BEA's recently released social computing products: Stephen Morais, architect for BEA AquaLogic Pages, Joseph Stanko, architect for BEA AquaLogic Ensemble, and Dax Farhang, product manager for BEA AquaLogic Pathways.
Audio TranscriptAjay Gandhi: Although the definition of Web 2.0 is still evolving, most people agree that the term refers to empowering every user to become an active participant in creating, sharing, filtering, and remixing all sorts of information on the Web, from instant messages and blogs to YouTube videos and virtual worlds. More recently, there has been a groundswell of interest in leveraging Web 2.0 technologies, design patterns, and usage methodologies for the enterprise. People have focused on user-oriented tools such as wikis, blogs, tagging, and simple mashup platforms—but not on the tough questions about architecting Web 2.0 applications and the implications for an existing IT stack, governance, data management, and development lifecycles. But applications leveraging Web 2.0 for the enterprise will potentially require a different interplay between business users, IT developers, and architects. Can you give us your view on how business/IT alignment will change as Web 2.0 becomes more prevalent in the enterprise? Stephen Morais: Web 2.0 means more end-user content, which is a good thing. But it may impose an additional burden on IT, similar to the problem of "rogue" access to databases. They key is to give business the tools they need, but in a way that allows for centralized management and governance. Joseph Stanko: Today the relationship between business and IT is adversarial: Business often sees IT as a bottleneck, and IT complains about being asked to meet greater demands on smaller budgets. I believe Web 2.0 technologies can start to make business more of a participant and a partner in IT projects. If business users can help build part of the app, they are more likely to get what they want, and power users may even build it themselves. Business users can become active "prosumers" rather than just consumers of IT. Dax Farhang: Web 2.0 will help IT keep up with the dynamic needs of the business. Web 2.0 will support a better balance by enabling organizations to create applications in a cost-effective manner while still working within a framework that is owned and maintained by IT. Gandhi: Most enterprises use portal technologies as a primary end-user interface positioned in the user interaction layer of an IT architecture. How do participant-oriented Web 2.0 tools complement and/or overlap this existing infrastructure? Stanko: The SOA infrastructure provides security, content management, data sources, search, and so on for both Web 2.0 and portals. So I don't see this as two competing infrastructures but two different types of applications built on a common infrastructure. Farhang: Portals aggregate data or knowledge from a number of disparate systems. Web 2.0 represents the aggregation of the people that own or use the data. Portals represent the unified voice of the systems, and Web 2.0 represents the unified voice of the people; they are two halves of the whole. Portals represent well thought-out IT initiatives while Web 2.0 typically involves situational applications that may or may not have a long lifetime. Both can enhance the enterprise's ability to meet both long-term and short-term needs. Gandhi: Many large enterprises have significant SOA initiatives underway. Do these social computing products derail SOA or is there a win-win possibility? Morais: SOA neither precludes nor implies social computing, and vice versa. That said, the more service-oriented your architecture, the easier it is to create the types of mashups that are characteristic of certain Web 2.0 apps. Stanko: The real value that SOA promises is reuse. Ultimately, social computing enhances this by allowing power users and knowledge workers to reuse services to build situational apps. Farhang: Governance is an area where SOA and Web 2.0 may intersect. SOA initiatives help enterprises catalog the services that are available across the IT infrastructure. Social computing applications such as mashups help the enterprise leverage these catalogued services, while turnkey applications and data sharing may create a need for data governance. Gandhi: Great point. A big part of SOA is governance whereas Web 2.0 is all about free-form collaboration, sharing, and transparency. How should enterprise architects balance Web 2.0 "openness" and the need for IT governance? Farhang: Roles, capabilities, and security are definitely the key. People usually know what data should or should not be shared, but before Web 2.0 they lacked effective mechanisms for doing so. By delegating the right privileges to the right individuals and allowing them to define the collaborative body based on their understanding, enterprises can strike the right balance between governance and openness. Morais: Security and governance are important for, say, a payroll system, but probably less so for ad hoc collaboration between and within groups. Governance has to be unobtrusive. If the burden of governance makes people shy way from using the technology, you're shooting yourself in the foot. Stanko: Accountability is key, too. With Ensemble, for example, you can easily add usage analytics and audit trails to Web apps and mashups in your enterprise. It's a myth that governance isn't a concern for the Web 2.0 sites on the consumer Web. For example, people deface Wikipedia, and many controversial subjects are locked down. The right infrastructure gives you flexibility to lock down sensitive information without locking people out of the technology. Gandhi: Corporate knowledge workers are increasingly using tools to build Web 2.0-style mashup apps that combine enterprise data, ad hoc user-generated content, and content from external Internet services. Where does this end-user development fit into a company's portal and SOA-based composite application development approach? Morais: It adds another "flavor" of service interfaces to the mix. While Java and SOAP APIs are great for developers, they're too complex for end users. They need something simpler with a higher level of abstraction. REST and RSS are good examples. Also important are visual tools that end users can use to tie these things together—tools like BEA AquaLogic Pages that simplify the underlying complexity of the interfaces but still allow users to get at the data. This approach will require a better separation of roles between what IT will focus on versus end users. Stanko: There are shorter time frames for mashups and other Web 2.0 applications compared to typical application development cycles. End users will develop new apps on the spur of the moment to meet their needs, and a small percentage of those apps will get promoted and require IT to deploy them on an enterprise scale. Gandhi: A key part of a mashup's value is combining enterprise data with other information. How can enterprises expose corporate data to participative tools where end users can share data and potentially even modify it? Morais: Start simple. RSS is probably a great way to do this. There are scores of RSS consumption tools out there for end users. And 80 percent of the corporate data that users would care about for a mashup would lend itself to an RSS feed-based approach. Stanko: Adding an RSS feed to a blog or a news site doesn't give me as a reader of that site access to anything new, but it does give me new flexibility in how I use that information. This supports employees' mobile lifestyles and helps address information overload. Farhang: Companies already have an understanding of the data that their users are utilizing on a day-to-day basis. By putting that data in a more digestible format, such as RSS, it can easily be leveraged within Web 2.0 applications. Gandhi: Many enterprise IT shops rely on standards, especially key vendor-endorsed standards to guide their product and deployment decisions. Does Web 2.0 enter in the standards discussion, especially technologies such as REST and RSS? Morais: REST and RSS are definitely part of the discussion. They have a lot of overlap—you don't need two different APIs, because REST and RSS are essentially the same, RSS just being a type of REST. Stanko: Some of the standards that were written by big committees have ended up getting in the way of simple, elegant software. A great example of this is the WSRP standard. It's far more complex than it needs to be. This has hindered its adoption. No one on the consumer Web uses it—they're using a much more RESTful approach with a simpler design that uses widgets and gadgets. Farhang: Standards are critical for the success of Web 2.0 in the enterprise. In order to leverage internal and external data sources in Web 2.0 applications, IT needs standard methods for how the data is made available to the applications and users that are consuming it. Gandhi: How is the emergence of Web 2.0 affecting application development in the enterprise? How must IT change to effectively incorporate this new app development approach within existing software development lifecycle processes? Morais: IT needs to expose data at higher levels of abstraction through simple protocols like RSS and REST. IT also has to realize that unless you give people tools that meet their needs, they will find a way around you with rogue wikis and other applications. You need to empower users with tools that meet their business needs—and that also help provide the governance that your organization requires. Farhang: I think there are two ways that Web 2.0 application development will affect the traditional software development lifecycle. First, IT needs to move away from the notion of building end-to-end applications and instead needs to embrace the notion of building more widgets or toolsets for the end users who are building the situational applications. Also, IT needs to embrace a more iterative approach to development, rapidly developing widgets and prototypes and delivering them to end users for feedback. This ultimately means a faster time to market for end-user created applications. Stanko: Web 2.0 application development also means rapid creation of lots of user-generated content, and people collaborating on a daily basis. You can't bend that into an existing development lifecycle process, because the timeframes don't match. Plus, your full-blown development processes aren't required for small applications designed for immediate and short-term deployment to a small group of users. It's overkill—and it gets in the way of rapidly deploying those applications. References
Ajay Gandhi is the Director for Enterprise Social Computing at BEA Systems where he heads the worldwide strategy, partnership and marketing initiatives for Web 2.0 social computing products within BEA's Business Interaction Division. Return to Arch2Arch. 
|
Podcast Tools Bookmark Podcast
|
