SOA Architecture, Governance, and Industry Standards in the Enterprise

Paul Lipton

Subscribe to Paul Lipton: eMailAlertsEmail Alerts
Get Paul Lipton: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn


Related Topics: SOA & WOA Magazine

SOA & WOA: Article

Composition and Management of Web Services

Development that makes a good symphony

Father James Keller, the founder of a religious order called The Christophers and a popular religious television and radio personality in his day, wrote an amusing story about Mozart. He said, "A young man, just beginning the study of musical composition, once went to Mozart and asked him the formula for developing the theme of a symphony. Mozart suggested that a symphony was rather an ambitious project for a beginner; perhaps the young man might better try his hand at something simpler first. 'But you were writing symphonies when you were my age' the student protested. 'Yes, but I didn't have to ask how.'"

Mozart was able to compose monumental symphonies out of simple melodic themes. Moreover, Mozart's symphonies always appear to work well as a whole. If there are bugs or mistakes in his compositions, it is usually far beyond most people's abilities to distinguish them. For those of us lacking a Mozart-like ability to create near-magical distributed computing solutions of perfect reliability or robustness, the idea of a service-oriented architecture (SOA) implemented using Web services standards could not have come at a better time.

Service-Oriented Architecture
A good way to understand the importance of an SOA is to think about classical chamber music. In a chamber group there is no conductor. With only a few instruments and frequently no percussion, the cello often serves as the foundation for higher-level musical activity. In short, its deep tones carry the beat. A service-oriented architecture is frequently a foundation for higher-level activity and offers many benefits. Used correctly, it can enable the creation of higher-level distributed applications based on Web services that are less brittle than many systems developed in the past based on technologies like Microsoft DCOM or Java RMI.

One of the key concepts behind an SOA is the idea of loose coupling - of reducing the dependency between distributed components through the dynamic discovery of services. This is an excellent architectural approach to take for many applications. Rather than hard coding dependencies (tight coupling) between components, loose coupling allows distributed applications to be less brittle and more responsive to changing business conditions. Services that provide specific functionality can more easily be switched with equivalent services and alternative services more easily invoked without breaking the application. Increasingly, Web services are the choice for implementing a SOA within and across enterprises.

Web Services in an SOA
Today Web services are commonly understood to mean a collection of standards, derived from XML, that facilitate interoperability between diverse platforms and applications. Web services have found increasing success and use in a broad range of companies from e-retailers like Amazon to financial giants like Merrill Lynch. Smaller companies are also finding the affordability and rapid return on investment (ROI) of Web services very compelling. As a result of this industry momentum, most informed observers these days believe that Web services will increasingly be used as the foundation for building new distributed applications both within the corporate firewall and across enterprise domains as part of increasingly complex and mission-critical B2B and electronic marketplace scenarios.

The relationship between a SOA and Web services is itself loose. It is possible to create a SOA without using Web services, but the core Web service standards such as SOAP for message protocols, WSDL for service description, and UDDI for service lookup offer the best mechanism to implement a SOA across platforms, development languages, and enterprise domains. There are many reasons for this, but the fact that these core standards are based on XML is certainly an important factor. XML's extensibility, platform neutrality, self-description capabilities, and more have all been leveraged by the core Web services standards as well as the newer Web services standards, such as OASIS WSBPEL and WSS (a continuation of the work done on the WS-Security specification).

Composable Web Services Standards
But there is another aspect of Web services standards that has significantly contributed to their success. Web services standards have largely been designed to be composable. Composability is a design principle that stipulates that it is possible to build more powerful and feature-rich functionality from simpler elements. Each simple element must have discrete, independent capabilities of its own, but can also be added to (composed with) other elements to create more complex solutions.

Web services composability works along similar lines. Again, thanks to the extensibility of XML, it is possible to compose additional independent functionality into any SOAP message or other types of Web services document, such as WSDL. Listing 1 illustrates a simplified example of this. The code shows that it is possible to add WS-Security information and/or WS-Reliability information (one of the specifications contributed to the OASIS Web Services Reliable Messaging (WS-RM) technical committee) to a SOAP message without changing any other information in the message and without impacting consumers of the SOAP message that may not be able to process that additional information. Similarly, it is possible to add WS-Policy information to a WSDL document without interfering or intermingling with the standard description information that is typically found in WSDL.

Consumers of a Web service can take an incremental approach to utilizing information in a Web services document (normally a SOAP message), rather than being forced to understand it all. The creators of a service consumer must only support those standards that are necessary for their business requirements. For example, if they need message reliability added to SOAP, they can implement support for handling the WS-RM elements that could be placed in the SOAP messages. The ability to pick and choose just what you need for your business requirements helps reduce the complexity of service implementation, and therefore the cost.

Composable Web Services Standards in a SOA
One of the benefits of a SOA is that it can encourage the reuse of services, enabling many service consumers to use the same service in order to fulfill business and technical requirements. This can reduce development costs and speed time to market. In an SOA it is easier to locate and share services because there is a central repository that contains service information that can be utilized by service consumers. The Web services UDDI standard improves upon this by providing a universally accepted service discovery and publication mechanism. Because UDDI is so widely accepted (there are alternatives, but none likely to figure prominently in most solutions), there is a plethora of low-cost and easy-to-use development environments able to utilize UDDI and other Web services standards that are available from a variety of vendors, as well as the open source community.

This same incremental approach to utilizing information in a Web services document can also be of immense benefit in a SOA. Consumers need to use only those standards composed within the service that they actually need, while other consumers are free to use the service along with a different mix of composed standards that are useful to them. So, Web services standards composability not only increases the usefulness of a service- oriented architecture, but also facilitates loose coupling in that architecture as services do not have to rigidly match the requirements of only one consumer or one group of homogeneous consumers. Rather, consumers can be diverse, with different requirements, and yet still interoperate with the same service.

Management of Web Services in a SOA
Clearly, the advantages of a SOA based on Web services standards are driving ever-increasing industry momentum and IT commitment. This trend is already beginning to push the use of Web services beyond the corporate firewall toward new B2B scenarios and electronic marketplace opportunities. As these opportunities grow in value and complexity, the need to meet service obligations and to ensure the reliability and performance of Web services that are often themselves utilized by higher-level, mission-critical Web services becomes paramount. Yet this can be a daunting and near-impossible task without management solutions that are specific to Web services, functioning at the level of the SOAP messages that form the basis of an SOA in action. In other words, SOAP message traffic exists at a logical level above the middleware, operating systems, and hardware that forms the supporting infrastructure for Web services. It is the semantic content of those SOAP messages, including both business and security information, that must be observed in order to properly manage a SOA and its supporting Web services.

Another consideration is that well-managed enterprises often have existing enterprise management solutions from leading vendors like CA. These solutions typically offer comprehensive IT infrastructure management from the hardware to the operating systems and middleware. There is a need to unify management at both the Web services and the infrastructure level in a consistent fashion and using a consistent user interface.

Such unification is needed for two reasons. First, a Web services management solution that is not unified with IT infrastructure management can add to the cost, and reduce the responsiveness, of operations staff to service problems. This is because the IT operations staff is already familiar with the GUI and metaphors expressed by their existing enterprise management solution. To appropriately monitor and control Web services in a SOA, operations personnel would either have to be trained on both systems or learn to live with the overhead of miscommunication that would inevitably be caused by the dissimilar terminology and user interfaces in the two management systems.

Of even greater importance is the extreme difficulty in uncovering the true root cause of service problems when the integration between Web services management and IT infrastructure management is superficial or absent. At least for those services deployed within the enterprise domain, there must be some means of correlating a service with its underlying IT infrastructure. For example, service A may be performing poorly (see Figure 1). However, that poor performance may be the result of problems with other services that are consumed by service A. Web services management software should be able to help determine which service, even in complex environments, is the actual cause of service disruption in a SOA. In this example, let's assume that slow response time from service B is affecting service A. If service B was deployed in the domain of a business partner, it might be sufficient to notify the partner of the problem. However, if service B is deployed on your own IT infrastructure, then significant and meaningful integration with the enterprise management system is required in order to correlate the problem of service B with any possible issues with the underlying IT infrastructure. A unified approach to IT infrastructure and Web services management would allow us to determine the exact underlying cause of service B's poor performance, which in this case might be database memory problems.

 

Enterprise management also plays an important role in rectifying problems in the underlying IT infrastructure. With advanced on-demand capabilities it is often possible to rectify even serious resource allocation problems automatically. This sort of rapid response is essential in the fluid environment of an SOA. So, as Web services evolve toward more mission-critical scenarios, a unified approach to enterprise and Web services management will become an increasingly important requirement. Similarly, requirements for meaningful integration with enterprise security systems will also be a long-term trend for Web services management systems. This integration is likely to go far beyond superficial lookup of identity towards unified administrative portals and other advanced features.

A Web Services Management Standard
Although many management functions require SOAP message observation, especially when meaning or context is a concern, many management functions would be better defined as a management capability of the Web service itself (or its proxy or intermediary). For example, useful management behaviors defined for Web services would likely include life-cycle support and notification for a variety of events (such as a new version deployed or a changed interface), state management (start/stop/up/down/ overloaded /etc.) to monitor or change the behavior or state of the service, configuration management, and much more.

By defining standard management capabilities and interface descriptions, a Web service or its containing platform or proxy could provide management interfaces that would be usable by any management software. Such a standard would allow both platform vendors and Web service developers to provide manageability to Web services management systems that would otherwise be proprietary and platform specific. For example, an application server could automatically compose and deploy standard management interfaces into any Web service.

Since an important benefit of Web services is interoperability between dissimilar systems and applications, SOAs often contain diverse Web services platforms. A standard would enable Web services management systems to provide deeper monitoring and control capabilities of all Web services platforms that comply with the standard. It would even be possible to provide some or all of this manageability across administrative domains; this would be of value in many scenarios such as B2B relationships between trusted partners.

The Web services community has long been aware of the advantages such a standard could offer. Work was begun in the Management Task Force subcommittee of the W3C Web Services Architecture Working Group, but as the importance and extent of the work required became better understood, that work was moved to a new technical committee (TC) formed under the OASIS organization in March of 2003. That TC, OASIS WSDM (Web Services Distributed Management), is composed of a diverse group of companies interested in management issues, including leading management software vendors such as CA, HP, and IBM. The TC charter could be summarized as carrying on with the initial work of the Management Task Force with wider scope, working towards precise requirements definitions and specifications.

It is interesting to note that in addition to the management of Web services, referred to in the TC as MOWS, the TC is pursuing the requirements definition and specification of the management of IT resources, such as disk drives, operating systems, and routers using Web services. This effort is referred to in the TC as Management Using Web Services, or MUWS. While such an effort might be surprising, given that management of these types of IT resources is already possible using other standards, such as CIM, there is little doubt that the effort is worthwhile. Since Web services technology could potentially offer the same benefits to management software vendors as it does to business, it could be used to create more manageable and interoperable IT resources that would work better across multiple management software vendors, topologies, networks, and platforms. MUWS also forms the common ground for much of the work on MOWS, which is simply the more specific case with additional Web services specifics. The requirements definition for both MUWS and MOWS is essentially complete, and I believe we will see a finished specification from the TC by the end of the first calendar quarter of 2004.

Composition and Web Services Management
The final specification of the OASIS WSDM TC is almost certain to support the concept of composition discussed above. The formal WSDM specification was not available at this writing. So, in order to illustrate how composition can be used to extend Web services management, we will use WS-Manageability, a key Web services management specification that was donated to the OASIS WSDM TC by management software vendors CA, IBM, and Talking Blocks (recently acquired by HP).

Although a detailed description of WS-Manageability is beyond the scope of this article, the beauty of this specification is that it already represents a nonproprietary view of Web services management since it is the result of collaboration and compromise between three major vendors. It is a simple and easy-to-understand specification that adheres very strongly to the philosophy of Web services and SOA by limiting itself to the definition of the management capability of a manageable Web service and the specification of its management interfaces. In other words, WS-Manageability does not dictate the implementation of management software. It simply specifies additional Web services interfaces and their semantics that can be composed into a Web service's description to add management capability to that Web service.

This pure service-oriented approach to Web services management allows manageability to be expressed using virtually any service description standard. In Web services, service description is typically the province of WSDL (Web Services Description Language), but other important technologies use other standards. For example, grid computing is a technology based on Web services that is just starting to bloom. It provides many exciting potential benefits in areas of business computing especially those tasks that require large- scale computation. The Global Grid Forum (GGF) leads the standardization efforts in this area. Subgroups within GGF have defined an extended version of WSDL called GWSDL for expressing the capabilities of services in a grid environment. WS-Manageability's neutral and flexible approach to Web services management allowed the authors to easily provide descriptions of Web services management interfaces in both WSDL and GWSDL. Similarly, it should be possible to express WS-Manageability to many other models, and to express additional management information in WSDL using specifications such as WS-Policy.

Composability also applies to the actual SOAP messages that will be sent back and forth from management services. Clearly, Web services specifications that are already available or currently in development could be used to extend the management capabilities of SOAs built with Web services by providing a wide variety of functionality beyond the OASIS WSDM specification itself, such as reliable messaging, asynchronous message delivery, security, transactions, and more. Composition would allow Web services platform vendors and management software vendors to provide more advanced management architectures, as needed, while still supporting simpler management solutions.

Conclusion
An important but less obvious benefit of Web services is composability - a benefit that is derived from XML and carefully nurtured by the Web services community. This ability of Web services to incrementally enable additional independent functionality, as needed, maximizes the usefulness of Web services in building an SOA, which is an important goal for many enterprises.

Web services management standards like OASIS WSDM are absolutely crucial to the long-term success of Web services, especially given the clear trend toward more complex B2B and electronic marketplaces based on a service-oriented architecture. The donation of WS-Manageability by CA, IBM, and Talking Blocks to the OASIS WSDM TC is an excellent example of the continued awareness and dedication on the part of major management software vendors toward supporting composable, open Web services specifications as the cornerstone of a manageable SOA.

The Web services community, although sometimes filled with discordant notes, nonetheless is successfully composing a set of standards and solutions of symphonic stature. Although the progress of Web services standards often feels closer to the clashing rhythms of pyrotechnical 20th-century works like Stravinsky's "Rite of Spring" than to peaceful and bucolic works such as Beethoven's 6th symphony, the Web services community is methodically and successfully composing a useful and intriguing work of distributed technology that will have enduring value.

More Stories By Paul Lipton

Paul Lipton is VP of Industry Standards and Open Source at CA Technologies. He coordinates CA Technologies’ strategy and participation in those areas while also functioning as part of CA Labs. He is co-chair of the OASIS TOSCA Technical Committee, and also serves on the Board of Directors of the open source Eclipse Foundation, as well as both the Object Management Group and the Distributed Management Task Force in addition to other significant technical and leadership roles in many leading industry organizations such as the OASIS, W3C and INCITS.

Lipton is also an approved US delegate to the international standards organization ISO, as a member of the subcommittee focused on international cloud standards. He is a founding member of the CA Council for Technical Excellence where he leads a team focused on emerging technologies, a Java Champion, and Microsoft MVP.

Comments (0)

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.