Monitoring an SOA

Monitoring an SOA is essential to ensuring reliable operations and resolving problems quickly, in order to deliver customer service at agree-upon levels. However, the very nature of an SOA makes monitoring it difficult.

While enterprise applications exist in "silos," with a business process executing on a single platform that can be monitored from the bottom up, the component services that make up an SOA business process execute across distributed, heterogeneous systems. As a result, they can't be adequately monitored by traditional application infrastructure management tools.

Furthermore, performance requirements for a single service may vary depending on the service consumer. This variety of requirements multiplies when a service is shared by different business processes. Standard management tools, which report on average system response times, can't provide information necessary to understand how an individual SOA transaction, or any service within it, is performing or if there is a problem.

Monitoring an SOA: Unique Capabilities

As a result, monitoring an SOA requires visibility into the performance of each service as it executes in each individual SOA transaction—across distributed, heterogeneous systems. Only with this end-to-end visibility can SOA administrators manage an SOA and spot problems quickly, to ensure reliable operations.

To achieve this level of visibility, in turn, requires the ability to discover all services executing in the runtime environment and, then, to monitor or follow their interaction throughout an SOA transaction.

Monitoring an SOA with Progress Actional

Using unique, patent-pending technologies, Progress® Actional® for SOA Operations provides the end-to-end visibility into SOA operation required to monitor SOA performance and availability and to spot performance problems quickly, in other words, to ensure reliable operations.

The product includes agents, or points of visibility, that monitor the individual SOA components in the SOA network as well as a central SOA management server that maintains the repository of service-level policies and distributes them to the agents. The Actional server also provides summarized data from the runtime environment to operations staff through a graphical, browser-based interface, enabling them to understand and manage SOA operations.

These capabilities provide:

• Automatic discovery of runtime services and their dependencies. Once an agent is installed on an SOA component (e.g., an application server, network device, database, etc.), the agent discovers each service used or invoked by that component as well as its interactions with non-instrumented components. An agent can also detect different kinds of services, not only Web services. As a result, Actional agents discover all services and their dependencies in an end-to-end SOA transaction, which they communicate to the SOA management server.
  
  By discovering all SOA components and their dependencies, the agents enable SOA monitoring. They also detect rogue services, which can be overloading SOA system capacity or even violating security and compliance policies (for example, exposing sensitive information to unauthorized parties). Just as important, these agents perform discovery and monitoring without programming, application modification, or pre-discovery service configuration.
  
  
• Non-intrusive runtime monitoring. Actional for SOA operations monitors SOA traffic as follows. Each agent on an SOA network node includes interceptors for each component monitored plus an analyzer. Interceptors copy the minimum data needed to monitor their respective components from on-the-fly messages and function calls to an in-memory buffer—in just 10s of microseconds, causing minimal application processing latency.
  
  The analyzer processes the collected data, tests for service-level policy violations and communicates any violations to the Actional server as alerts. It also collects aggregate statistics and sends them to the server on a user-specified basis, every few minutes.
  
  The interceptors also add a non-impacting tracer to transactions that are subject to policies, collect and correlate tracer information, and then communicate it to the Actional server, enabling transaction tracing across components. The result is end-to-end transaction tracing, made visual in an Actional flow map.

By providing this end-to-end visibility across the SOA network, these two capabilities supply the basis for managing SOA operations, including the ability to quickly trace SOA operations issues back to their root cause.

For More Information on Monitoring SOA Operations

For an in-depth understanding of how automatic service discovery and monitoring work, read the whitepaper: Achieving SOA Operations Excellence with Progress Actional.