"There must be millions of people," wrote columnist Robert Benchley about 70 years ago, "who are no more equipped than I am to guide a motor vehicle through any more of an emergency than a sudden light breeze. The logical ending to the whole situation is for all the automobiles in the world to pile up on top of one another at one big cross-road."
When people talk about an Information Superhighway, Benchleys image quickly comes to my mind. In the same way that Benchley could never have imagined an H2 bearing down on a Mini, the people who built the Internet could never have imagined zombie bot nets mounting distributed-denial-of-service attacks on Net-edge cache servers. The Internet was built to tolerate random failures, not to withstand deliberate and focused attacks; it seems to me that new Internet initiatives still tend toward a science-project definition of technical success that says, "once it can be shown to work, its done."
Web services are definitely being driven, so to speak, as if they were past the prototype point, with real companies betting real projects on service-oriented frameworks such as Microsofts .Net and Java 2 Platform, Enterprise Edition. Both architectures have their success stories; both have their strengths and their weaknesses, with developer strength on one platform or another being as good a reason as any to stick with it.
Its typical, though, to find that an early version of a vehicle sometimes demands a few extra skills. Like Benchleys 1930s auto drivers, who had to learn to use a stick shift as the price of mobility, todays Web service developers have to shift for themselves in providing security mechanisms that will at some point become well-integrated into protocols and development tools. It would be wrong to call Web services too immature to use, but it would also be careless to ignore their current weak points when either designing or costing a project.
Nor do Web service standards bodies have the luxury of aiming at a stationary target, as they seek to drive the service model toward a destination of credible security. Security algorithms are under the same kind of scrutiny in public, and very likely attack in private, as any other link in the chain of trust: New research results suggest that security algorithms with respected pedigrees may harbor hidden weaknesses.
At some point, being a driver stopped demanding mechanical skills; at some point, technical support for cars could be bought at any downtown intersection. Likewise, security is on its way toward a model of road maps for sale. After years on the defensive, IT vendors are even close to the point of claiming security as a basis for strategic differentiation.
In the meantime, though, Ill drive with extra care when I find myself in heavy Net traffic—and the wind begins to blow.
Tell me whats buffeting your development plans at firstname.lastname@example.org.