In a recent SDTimes column, David Linthicum discussed "mashups" and whether they are consistent with SOA principles. David identifies two broad categories of mashups: visual mashups like Google maps displaying locations from a sex offender database, and non-visual mashups combining two or more other services for consumption by another application or service. Non-visual mashups are exactly what I've discussed in the past as "data services mashups" (see this JavaOne interview with SearchSOA ). In fact, XAware is all about non-visual mashups, as its main benefit is the ability to pull data sets from multiple sources, and expose the result as a higher level service.

So, are mashups consistent with SOA? David unequivocally says that they are, and claims that they are likely the most successful example of SOA that we've seen to date. David shows that mashups exhibit many core characteristics that SOA attempts to achieve:

Though mashups did not emerge from the core concepts of SOA, they indeed provide some core SOA mechanisms, including:

1. The ability to place volatility into a single domain, thus allowing for changes and for agility.
2. The ability to leverage services, both for information and behavior.
3. The ability to bind together many back-end systems, making new and innovative uses of those systems.

In addition, mashups share the important characteristic of "thriftiness" with service oriented computing. Wherever possible, existing resources are reused in new ways, in contrast to purchasing replacement systems or capabilities. Ultimately, mashups provide agility to a business, by quickly providing capabilities as new services supporting new business functions. This, more than anything, is convincingly service-oriented.