I believe that Web services may make the long-standing battle of open source software (OSS) versus closed source software (CSS) almost irrelevant. What!, you cry in incredulity. He can't mean that! How can he discount the efforts and thinking of some of the world's most brilliant programmers? However, before you blast out a scathing e-mail flaming me to a crisp, listen to my reasons, then decide if you don't agree with me.

The Middle-Mile Solution
Recently I've come to think of Web services as a "Middle-Mile" solution. To understand why, it's helpful to look at a simple example - the steaming cup of tea on my desk. The essential players in getting my tea in the morning are the grower/manufacturer, the distributor, the local store, and, of course, me - the end user. The transportation of the tea leaves or "raw goods" from the manufacturer to the distributor is the figurative "First Mile." The distance from the distributor to the local store is the "Middle Mile," and my walk to the store to get my tea leaves is the "Last Mile." The trains and trucks that form this middle mile by transporting tea leaves to my local store rely on customers like me to consume the tea leaves when we come into the store to buy the tea. Likewise, I rely on them to transport tea leaves to my local store so they're ready and waiting when I want tea.

Web services provide the solution equivalent to the trucks and trains in my tea example - they move the goods from the manufacturer/distributor to the local store/end user. However, rather than a physical "hard good" such as tea leaves, Web services instead transport an electronic "soft good." In this way, Web services are the middle-mile solution for soft goods - dealing with bits, not atoms.

These electronic soft goods are usually small pieces of a business's core logic, exposed so they can be called by a business partner, customer, employee, or another application. Almost every business has discrete shareable business functions, ones that are often core to a particular business interaction with its partners. In addition, almost every business wants to use shareable business functions from their partners, to exchange information surrounding commerce. What's needed is agreement on the actual mechanism and standards - enter Web services.

Standards for the Infrastructure
Web services define standards for the infrastructure to remotely call discrete shareable business functions. This isn't new. We've had the ability to invoke remote objects since our applications got bigger and we started to spread them around - moving to three-tier and then to n-tier architectures. However, in the case of Web services, it's done with protocols and data representations that are ubiquitous and Internet-friendly. The wire protocol is usually HTTP over TCP/IP but others are allowed (such as SMTP); the network data representation is XML; the interface definition language is WSDL (an application of XML); and the registry for Web services is UDDI. All of these standards are blessed and espoused by a collection of vendors the likes of which we've never seen before, including IBM, Microsoft, SAP, Compaq, Ariba, and many more.

HTTP, WSDL xSchema, and UDDI are the current offerings but other protocols, data encodings, interface definitions, and registries could become popular and they should be easily pluggable. The developer requirements to call a Web service are made trivial by the use of a SOAP ORB. This is analogous in function to a CORBA ORB in that it allows the developer to call methods on remote objects as if they were local objects without regard for the underlying mechanism. This includes support for strongly typed objects into and out of the method calls, which implies support for standard and customizable data marshallers/de-marshallers.

Implementations of almost all of the basic technological requirements are already available as OSS projects, Apache SOAP for example. Additional portions of the "Web Services Stack" are currently under development in ongoing OSS efforts - Apache AXIS (a better SOAP ORB), WSDL, UDDI. These efforts allow value to be added from the tools front. These tools should strive to make it easy for developers to wire together services into business process flows, and then to call these flows - otherwise known as applications.

Web services enable these next generation applications, which are really composed of software accessible across the Internet by common protocols, and open standards. These applications travel the "middle mile" between the trading partners and allow meaningful e-business conversations, with or without human involvement. Like open source software, Web services rely on the "network effect" Eric S. Raymond (author of the seminal works on open source software development including The Cathedral & the Bazaar, "Homesteading the Noosphere," and "The Magic Cauldron") describes - however, in a slightly different way. While Web services can be used between applications inside the same corporate network, or across a LAN between divisions, their value is fully realized when connecting trading partners.

The developer metaphor for invoking a Web service is very simple in theory and practice. First there's the matter of generating a stub from the WSDL file that describes the service: where it is, what it requires, and what it returns. Once this stub is created the developer uses it to call the remote code. This stub can also be used in larger constructs to wire together whole modules, sub-systems, or indeed whole processes. All this can be irrespective of what language any of the subtended Web services are written in, what operating system they're running on, or even where they physically execute.

The real value of Web services is their black box functionality. They can be used across devices, cross-platform, and cross-vendor. All any developer has to know is the location to the WSDL file and they can use the logic embedded in the Web service. The developer specifically does not own the execution environment.

Open Source vs Closed Source
Now that we know about Web services, let's return to the question of open source software versus closed source software. Most of you are probably familiar with object-oriented programming and have built applications on these principles. Two camps of thinking have emerged: the "Open Source Software" camp and the "Closed Source Software" camp.

Raymond, in his article "Homesteading the Noosphere," outlines some of the breadth and depth of the greatly varying "hacker" opinions. Among the zealous hackers he says that attitudes range from, "Free software is my life! I exist to create useful, beautiful programs and information resources, and then give them away," to, "Yes, open source is OK sometimes. I play with it and respect people who build it." And from the anti-commercial hackers the sentiments range from "Commercial software is theft and hoarding. I write free software to end this evil," to "Commercial software is fine, as long as I get the source or it does what I want it to do."

In his recent work "The Magic Cauldron," Raymond stated that when developing software, the following discriminators push projects towards open source:

• Reliability/stability/scalability are critical.
• Correctness of design and implementation can't readily be verified by means other than independent peer review.
• Software is critical to the user's control of his/her business.
• Software establishes or enables a common computing and communications infrastructure.
• Key methods (or functional equivalents of them) are part of common engineering knowledge.

Web services allow us to call the most critical business functions to our business; however, these might reside on systems and infrastructure that are owned and operated by our business partners. Our partners own these mission-critical core business functions. So, if these services don't work when they're called, your natural response will be to find another partner, not to try to run the code yourself. After all, you can't do everything, that's why you have partners.

This business function will either be something quite simple and well understood like an order entry system or a product catalog, or it will be insanely complex and very proprietary. In the first case peer review isn't necessary; in the latter your partner isn't going to open the source - this is their core business value-add.

Applications built with various Web services will be critical to your business. However, they're owned by your partner, and so the selection of partner is critical - again open or closed source doesn't matter - if the partner isn't there, the service won't be there.

The entire Web services stack relies on a common communications infrastructure - the Internet. However, the implementation can be called irrespective of the computing platform used.

A common communications infrastructure is used between Web services - and much of this is already open source. But the code that's used by a Web service is a different matter. Even though the process of a given business function is well understood, it's not in your business or your sphere of competency, and therefore you may use someone else's service but probably wouldn't be inclined to run the code in-house. Indeed, one of the benefits of sharing Web services is in not owning the implementation, which means you have no knowledge of the source.

Conclusion
So, whether your business processes are coded in open source or closed source software seems almost irrelevant when using Web services. Web services de-couple systems and abstract their interfaces through XML and HTTP. As such, they're capable of bridging the acrimonious chasm between the operating system vendors and may be powerful enough to hurdle the rift between the open source and closed source communities. In the end, you don't have to know the implementation details of a particular business function. All you need to know is that it works. If it doesn't, it's your business partner's responsibility to fix it - or it's your option to find another partner.