Q
Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Monitoring metrics for performance and availability within an SOA application

What typical metrics are monitored for performance and availability within an SOA application?

In many ways, the metrics are the same as traditional performance and availability metrics – things like response time, volume of requests, fault rates, etc. However, what differs for an SOA is not what the metrics are, but where you measure them.

Consider the following example: Imagine you have a "customer lookup" service that retrieves customer information, and its average response time is 5.5 seconds, which is below your defined 8 second threshold. In the traditional non-shared model, you would consider the service as operating successfully.

But, now let's look at an SOA environment: an environment where this same service is shared by two different applications. One of the applications always looks up individual customers using simple, precise criteria. As such, when it makes requests of the "customer lookup" service, the average response time is 1s. The other application, however, typically does complex queries that result in multiple customer records at a time, and so the average response time of its requests to the "customer lookup" service is 10s. In this world, the average response time is still 5.5s, so have you succeeded in meeting the service level expectations? No, because one consumer is consistently having the service level threshold exceeded.

While, in a traditional environment, it's enough to measure metrics like response time per-service, in an SOA environment, you need to measure, track, and alert on these metrics per-service-consumer. That is, you need to track the metrics independently for each consumer of the service to assure that the service is meeting the necessary expectations. Whether the consumers are external customers with which you have contractual service level obligations, or the consumers are internal applications that have certain (potentially unwritten) expectations of service level, the need to measure and track them independently is the same. No consumer cares how well you are serving other consumers; all they care about is how well they are being served.

Dig Deeper on Topics Archive

Have a question for an expert?

Please add a title for your question

Get answers from a TechTarget expert on whatever's puzzling you.

You will be able to add details on the next page.

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.

-ADS BY GOOGLE

SearchSoftwareQuality

SearchAWS

SearchCloudComputing

TheServerSide.com

Close