SOA Meets Web 2.0 at SOA Executive Forum
Rohit Khare was the moderator of this panel. He opened by referring to Tim O'Reilly's definition of Web 2.0, which was quite helpful, because the concept of this panel, Web 2.0 and SOA runs the risk of being one poorly defined term "Mashed up" with another arguably poorly defined term. Some efforts to define SOA include the OASIS SOA Reference Model. Rohit is a founder of KnowNow, and research director at CommerceNet.
The speakers included:
* Moderator: Rohit Khare, Research Director, CommerceNet
* James McGovern, Enterprise Architect, The Hartford
* Lew Tucker, CTO of AppExchange, Salesforce.com
* John Yapaola, Vice President & GM, Americas, Kapow Technologies, Inc.
* Robert J. Glushko, Ph.D., School of Information at UC Berkely
One of the first topics or phrases addressed was the idea of "Mashups", the notion of these ad-hoc integrations. John Yapaola talked about the 80/20 rule ("Pareto principle") which suggests that 80% of services can be centralized into a core platform of reusable services--while 20% can be maintained in decentralized and developed in a composite application environment.
Yapaola cited Nucleus Research top five ROI drivers in the Enterprise being:
- Breadth (how many people touch it)
- Repeatability (how many times a day)
- Cost (expense relative to value)
- Collaboration (will employees need to communicate to use it) and
- Reusability (can the information/data be reused)
Robert Glushko from UC Berkeley introduced himself by saying that the real value is driven from the enablers--and that app developers are wrongly getting all the glory. He said that architects and platform developers deserve all of the glory--and that while mashup artists get the public glory, the platform architects and API designers are the real heros.
Lew Tucker said that the ultimate success comes from use--and cited the example of a mashup between salesforce.com and Skype that allowed instant one-click calling of salesforce contacts. So true to his evangelical roots, he suggested that the primary value in the platform could only be expressed by the user base. The more users, the more value. Clearly the value chain is well understood by Steve Ballmer's excitement about developers (link to Google Video), which is a clear notion that the platform supports "killer apps" (in the case of Web 2.0, "killer mashups", and in turn this becomes value in the hands of users.
James McGovern pointed out that there was value in "Community" and "Participation". The opportunity to think about how your services could be used or reused outside of a vertical is exciting. It takes a business user to understand how something like an insurance-oriented service might be integrated into a car dealership application. So he suggests that business users get involved and knowledgeable about the capabilities of the underlying platform in order to maximize the value of the emergent platform.
Community models such as Sun Microsystem's Java bug parade (public access to the bug database) and WikiPedia are referenced. Lew Tucker pointed out that critical mass is needed and scale is needed in order to leverage the economies of scale within the business. Lew pointed out that there was interest in "tagging" within Salesforce for example the idea of creating a "hot lead" tag. The notion of community feedback is really just another class of metadata which enables new consumption patterns for services. For example, if a user reccommends use of a service, that user's recommendation provides almost a new consumption pattern for consuming the underlying service. So community metadata including tags, reputation, referrals, reviews.
Robert Glushko talked about "microformats", this veered the conversation towards sematic web areas, but this topic is quite useful for dealing with the issue of semantics. Sematics is one of three areas highlighed by Joe McKendrick as areas where SOA needs work. Microformats are an alternative to heavyweight standards-body driven structures for semantics--and seem to be useful to provide a counterweight to "universal" standards which can tend to be verticalized. Lew Tucker pointed out that microformats could themselves be seen as a standard, and that going forward there problems are going to be things we have to live with--and that it's neccesary over time to incorporate user feedback into the evolution of the interfaces.
Rohit Khare brought out Ray Ozzie's description of the "Cut and Paste" web. This would be akin to a desktop form of mashup which could allow for user interaction to govern mashups. The closest analogue to this is the UNIX "pipe" capability, which is not quite an end user functionality--but enables some forms of script driven integration.
Overall, an intellectually stimulating topic and a very smart panel. The panel raised more problems and issues that it solved, but the SOA community will benefit in hearing about these perspectives and stimulating further discussion on these problems in these terms. Hopefully, the InfoQ community can discuss these topics and further the community's knowledge about these speculative topic areas.
Enterprise Web 2.0
Problems with Enterprise Web 2.0