Web Services is a term that is being used to define a set of technologies that exposes business functionality over the Web as a set of automated interfaces. These automated interfaces allow businesses to discover and bind to interfaces at run-time, supposedly minimizing the amount of static preparation that is needed by other integration technologies. The question is, "Do Web Services solve the business problem they propose to solve?"
The business problem in the case of Web Services is the integration of extranet services into larger applications. My belief is that Web Services do not offer any great advancement forward in solving extranet integration, which I hope to illustrate throughout this article. Let's start by further identifying the business problem and the components of Web Services that serve as a solution.
The first part of the business problem is discovering the available services that have been exposed by a particular organization. These services may be transactional, informational, or visual. A transactional service allows the user of the service to invoke a transaction, such as ordering a new computer system. An informational service allows users to query information, such as the status of an already placed order or to request product documentation. A visual service will respond by providing data that is designed for visual display, such as in a Web browser.
The business problem is defined as two-parts: (1) how can we locate and identify a service for use, and (2) once found, how can we incorporate that service as part of a larger extranet application. It is important to recognize that this is about extranet applications here. Web Services is a powerful, but
potentially heavyweight, service that may fill the needs of an extremely large population, but might be excessive for intranet integration purposes.
Allow me to further clarify this point. There are those who confuse the entire Web Services architecture with the portion of Web Services that satisfies the second part of the business problem (described by #2 above). Indeed, there is a need to leverage and use services in a loosely coupled manner for intranet application integration as well as for extranet applications, but there are more lightweight technologies that are available to assist in these endeavors. We will cover this further in the section on "Leveraging Web Services."
For now, we can identify the problem as attempting to incorporate existing extranet services into internal business processes for the purpose of automation and increased communications.
How To Discover A Service
To discover these types of services, a group of companies, led by Microsoft, Ariba, and IBM, created the Universal Description, Discovery, and Integration (UDDI) specification. This specification forms the basis of a registry that is designed to store service profiles by organization, and contains all the necessary information required to understand the intent and invoke a particular service. Note that not all of this information is intended for machine-to-machine communications. A service interface might be represented by a telephone number as well as by a TCP/IP port.
The UDDI specification focuses on the storage, organization, and architecture of the registry service, and not on the usage of the information that is contained within. UDDI leverages the Simple Object Access Protocol (SOAP) to define interfaces for publishing and inquiring about the information in the UDDI repository.
Leveraging Web Services
Initially, some human intervention will probably be required to locate and identify a Web service. This will probably occur during the design phase of a new or modified business process. This Web service will then be incorporated into an application that extends across the firewall during one or more steps of the business process to retrieve information associated with that process.
For example, an automobile manufacturer may want to automate their supply chain by extending across the Internet and retrieving order and inventory status from a supplier. This will probably take place as part of a larger planning process that also leverages information from internal ERP and other operational systems.
The next set of technologies that assists with this type of initiative will help companies with binding and executing remote services in an automated fashion. Again, this is where some confusion occurs regarding the purpose of the binding technology. In the case of Web Services, there is an accepted specification called Web Service Description Language (WSDL) that is an XML document that defines the inputs and outputs of a Web service, including the XML Schemas that should be used to create the input and output documents.
Each WSDL document contains both an abstract definition of the service and how that service binds to a particular network im-plementation and data format bindings. The following example is from the WSDL specification and defines two messages that will be used to create a virtual port and how that port binds to an HTTP/SOAP implementation.
<message name="GetLastTradePriceInput">
<part name="body" element="xsd1:TradePrice"/>
</message>
<message name="GetLastTradePriceOutput">
<part name="body" element="xsd1:TradePriceResult"/>
</message>
GetLastTradePriceInput and GetLastTrade
PriceOutput each describe a message whose body consists of either TradePrice or TradePriceResult document types.
<portType name="StockQuotePortType">
<operation name="GetLastTradePrice">
<input message="tns:GetLastTradePriceInput"/>
<output message="tns:GetLastTradePriceOutput"/>
</operation>
</portType>
In the above section, portType defines a virtual port that creates an interface called StockQuotePortType with an operation called GetLastTradePrice. GetLastTradePrice is analogous to a method on the StockQuote PortType interface for those familiar with this type of terminology. It also defines that the input to this method is a GetLastTrade PriceInput and the output is returned as GetLastTradePriceOutput.
The binding section in Listing 1 physically links the interface and the method described earlier to a specific implementation. In this case, the binding states that the service is implemented on the URL specified by the soapAction attribute of the soap:operation element and that it returns a SOAP message with a body conforming to the stockquote XML Schema.
This section of the WSDL document is
critical for users of the service to understand as it tells them how to create the SOAP message for delivery. Other such bindings exist for HTTP GET/POST and MIME.
<service name="StockQuoteService">
<documentation>My first service</documentation>
<port name="StockQuotePort" binding="tns:StockQuoteBinding">
<soap:address location="http://example.com/stockquote"/>
</port>
</service>
Finally, we name the service StockQuote
Service, which links it to the SOAP binding and the abstract definition of the StockQuotePort. Users of this service will send their SOAP messages to the URL defined by the soap:address element. By definition then we can tell that there are multiple URLs used to implement this service. The external service definition is identified by the service element and the internal service implementation is identified by the binding element.
In the above example, a significant portion
of the specification is dedicated to late binding efforts that allow a service definition
to be connected to many different implementations. While this could be used for the purposes of intranet application integration, there are many more assumptions that can be made about the environment in which that type of integration is being performed. For example, it may prove more advantageous to use IBM's MQ Series to connect two disparate applications and to use a SOAP message between them to provide the loose coupling desired without inheriting all of the overhead of the multiple layers discussed here, which are necessary for extranet binding to occur.
Where Is This Leading?
It is my belief that XML is so flexible and can be so easily molded to fit any problem space that it has led to people using XML in scenarios where it is actually debilitating to the application instead of an enhancement. I call this illogical need of engineers to use XML without solid reasoning the "Not Invented For XML" (NIFX) syndrome-a derivation of the well-known Not Invented Here (NIH) syndrome. Web Services in total is one example of NIFX.
Web Services, like many other extranet integration technologies, still suffers from many of the same hurdles as its predecessors, namely CORBA, DCOM, Java RMI, or DCE. XML does provide a more flexible data serialization mechanism to use in tandem with an extranet integration, but adds a new dimension of complexity to the overall application. For example, there are now multiple layers of parsing and unmarshaling that must occur before the actual implementation is invoked. Also, it is as strongly tied to the need for standardization of data formats as any of the preexisting extranet integration technologies.
Additionally, Web Services does not provide solutions for any unanswered questions business executives might have with regard to exposing functionality to its trading partners. Indeed, many of the functions that businesses have chosen to expose they already do or could have with existing Web application technologies, such as HTTP, HTML, and servlets/Active Server Pages. Also, the security model surrounding Web Services needs time to mature as we continue to find holes in Internet security implementations on a monthly basis. And, WSDL is simply a new incarnation of the Interface Definition Language (IDL) that was used by Microsoft's DCOM and CORBA.
Web Services does offer an alternative to invoking publicly exposed functionality. That is, users are no longer tied to "scraping" ever-changing HTML forms in order to integrate over the Web, but still gain the benefit of a document metaphor for integration over a pure application programming interface. However, the downside risk is that users are now susceptible to XML Schema versioning the development of XML document handlers for the return values.
Web Services will become prevalent because anything XML is perceived by business to be simpler to use and to write applications to process - not because it is a major improvement over other extranet integration technologies. It is my belief that the industry would have been better served by focusing on using XML as a serialization format between disparate existing extranet integration technologies, thus leveraging the maturity of those solutions while minimizing the disparities between them.
Author Bio
JP Morgenthal is CTO of iKimbo, a leading provider of enterprise instant communications solutions. He is also author of Enterprise Application Integration with XML and Java and a world-renowned expert on distributed computing.
jpm@ikimbo.com
Where Are Web Services Going?, by JP Morgenthal
WSJ Vol 01 Issue 00 - pg.7
Listing 1: Linking the interface to a specific implementation
<binding name="StockQuoteSoapBinding" type="tns:StockQuotePortType">
<soap:binding style="document"
transport="http://schemas.xmlsoap.org/soap/http"/>
<operation name="GetLastTradePrice">
<soap:operation
soapAction="http://example.com/GetLastTradePrice"/>
<input>
<soap:body use="literal" namespace=http://example.com/stockquote.xsd
encodingStyle="http://schemas.xmlsoap.org/soap/encoding/"/>
</input>
<output>
<soap:body use="literal"
namespace="http://example.com/stockquote.xsd"
encodingStyle="http://schemas.xmlsoap.org/soap/encoding/"/>
</output>
</operation>
</binding>
