Early implementations by current adopters are solving business problems today, but more important, vendors are looking at them to help identify future requirements. Many of these highlight the need for additional standards to ensure interoperability between the various technologies. Currently the Web services standards portfolio is rapidly increasing in size to match functionality requirements, yet the area of management remains vague and incomplete.

Web Services Standards Evolution
The evolution of standards for Web services has come in three phases. The first or "connection" phase involved laying out the core, baseline standards: the XML Schema, SOAP, WSDL, and UDDI. These standards - the building blocks of Web services - are now in place, through the efforts of such standards groups as W3C, WS-I, etc.

The second and perhaps most complex phase is focused on security and reliability. In this phase, WS-I is working on critical Web services specifications like XML Digital Signature, XML Encryption, HTTP-R, SAML (Security Assertion Markup Language), and XACML (eXtensible Access Control Markup Language). Additionally, groups such as OASIS, WS-I (Web Service Interoperability Organization), and others are all working to present additional guidelines around these specifications.

The third, or "enterprise," phase will address provisioning, transactions, workflow, and systems management. Unfortunately this phase is in the very early stages of discussion, despite being one of the primary concerns among enterprise organizations.

Managing WebSphere-Based Web Services:
What Does It Mean?
There are currently multiple interpretations of what the phrase "managing Web services" means. Some companies (i.e., Actional and Infravio) offer products labeled "Web services management" for the integration and distribution of Web services, ensuring version control, change management, provisioning, etc. Additionally, there are still two interpretations specific to performance and availability management within the Web services market.

Recently the OASIS group announced its OASIS Management Protocol Technical Committee. The intent of this committee is to facilitate distributed systems management over the Internet by sharing management information as a Web service and integrating various management tools such as network management, systems management, application and component management, etc. IT managers view this initiative as the next SNMP, offering a common language to communicate performance management information. This will likely be driven by companies that offer an enterprise console for management data (example: Managed Objects), and niche management tools that will adopt an inexpensive way of offering integration into many enterprise management products.

Critical to the success of managing Web services are the performance and availability of the infrastructures required to successfully deliver a Web service. Other than security, performance and availability are often the primary concerns among those companies considering Web services initiatives. Currently there are no standards in place for monitoring and managing the performance of Web services; however, many groups, including DMTF (Distributed Management Task Force), WS-I, W3C, OASIS, and others, are investigating and drafting requirements.

Management Requirements for Delivering Web Services
Like any "service-oriented" IT solution, the success of a Web services transaction depends on two separate infrastructures to be performing optimally. The service requester first depends upon the availability of the Web services registry. The requester then relies on the availability of the provider and the performance of the provider's environment. Finally, the requester is always dependent on the overall integrity of her/his own infrastructure.

Additionally, the service provider is responsible for ensuring that all of its services are available on demand. Any time a service is down, that can typically translate into lost dollars. The provider must also offer acceptable performance levels to prevent requesters (customers or partners) from changing to a competitor. While availability is an absolute requirement, performance levels will be what differentiate one provider from another and will serve as the basis for internal and external service level agreements.

However, as stated earlier, there are still no standards in place to enable the manageability of Web services applications and infrastructures.

Challenges for Existing Management Tools
There are many challenges facing traditional management tools when attempting to ensure performance and availability of Web services. When integrating separate tools to address the many different elements within the infrastructure, IT is still left without visibility into various critical components.

IT managers will often default to systems management tools for their infrastructure monitoring of Web servers, portal servers, Web application servers, and database servers. Network management tools will provide protocol monitoring for such things as SOAP, HTTPS, FTP, e-mail, and TCP/IP. Component management tools may provide metrics on the performance and availability of JSPs, servlets, EJBs, and JDBC connections. Yet while this approach may address some needs within an enterprise environment, there are a number of issues that remain, including:

Leveraging Existing Technology - JMX
Existing technologies can be leveraged to help fill the void of performance management for Web services. The JMX (Java Management Extensions) specification defines the architecture and API requirements that provide Java developers with the ability to implement and expose the management functions of distributed applications. Through the use of JMX, Java-based applications can be queried and monitored during runtime to expose performance and availability information within individual Java components. New management solutions optimized for Web services environments now provide the ability to communicate with Java applications using JMX.

Management of Web services environments is available today through the use of JMX. By monitoring a JMX-enabled SOAP runtime, a management application can consume many critical performance attributes within a Web services infrastructure, including the total number of Web services deployed and the total number of calls to all services combined. Additionally, data can be monitored from individual services as well, including successful invocations, the number of invocations per method, the number of failed invocations, and average response times.

To accomplish this requires only two specific monitoring solutions and a free developer's tool. To monitor the network, new solutions from companies like SMARTS, Concord Communications, and Micromuse offer scalable solutions designed for Internet environments, and provide the flexibility of standard protocols and APIs to allow for easier integration.

To enable the SOAP runtime with JMX, IBM offers the Web Services Tool Kit. This award-winning development tool is available free as a downloadable solution from IBM's alphaWorks emerging technologies Web site. Within version 3.2.2 of the Web Services Tool Kit is a JMX-based management Web service. "An update of the Web service management demonstration includes a new JMX-based management Web service that allows MBeans stored within a JMX MBean server to be accessed and manipulated from a remote console. This technology allows a single console to monitor the status of JMX-enabled services and components throughout an enterprise through an AXIS (SOAP) client interface."

The third, and most important, product needed to manage a Web services infrastructure is a management solution designed to monitor application server environments, including the operating system, Web applications, and components. Products are available on the market today that provide comprehensive management of performance and availability of application server environments, including CPU, memory, file system, JVM, servlets, EJBs, and methods. In addition to the provider's infrastructure, information can also be ascertained from the registry's response time using SOAP, HTTPS, and TCP/IP, as well as the performance of individual services such as response time and invocations.

This combined approach to proactively managing Web services provides users with the scalability and functionality needed to ensure services are performing at optimal levels. While there are many solutions on the market capable of providing monitoring of application server components, few can combine those features with the ability to monitor, threshold, and act upon any MBean within any JMX-enabled application, as well as the ability to monitor specific Java method calls for measuring response times or invocations. The ideal product for this type of environment will also provide this monitoring information in business-centric views, such as the health or success/failure ratio of a particular service, as well as views for both IT and development.

A Web services industry analyst firm, The Stencil Group, in their report, "Understanding Web Services Management," summarizes the need for management with functional elements for services management solutions. Within the category of runtime monitoring and control, they suggest that companies look for alerts and exception events, heartbeat (continual monitoring of availability), performance metering, and service prioritization. Despite the lack of standards pertaining to management of Web services for WebSphere environments, leveraging JMX and the combined solutions mentioned above addresses each of these needs today, while providing additional detail to prevent service failure or at least immediately determine the probable cause of issues and notify the appropriate users.

Resources