Enterprise Services & Managing Quality of Service

[Former Member]

Hello,

the documents published recently regarding ESA are pretty much aligned with the general idea to follow the idea of SOA. We talk a lot about enterprise level functionality and the reuse of high-level patterns being developed. However there is not much talking (yet) about the managing and monitoring of those enterprise services. Eg. Enterprise Services without well defined service-level-agreements attached to them and a platform supporting to manage&monitor (quality of service) these services tend to be applied just like any other application-service.

The question raising here: Where (based on what technology/platform) does SAP want to manage such SLAs allowing also business managers to understand what quality of service is being provided to their business functions. So far I haven´t spotted anything supporting that layer...(in the sense of "Click on an enterprise service, eg. define alert if response is greater than x sec or n service call failed, category of enterprise service...)

Regards

Rudi

[david_sonnenschein]

Rudi,

It is never any fun to say "It depends". But that is the way I feel I have to start this response. I think your question does hit the next level of required detail in an ESA beyond the exposure of services to the enterprise. Immediately after exposing services, basic SLA's and monitoring becomes critical.

There are in fact monitoring capabilities in XI 3.0 that should give us the first round of basic "Service Management". XI also offers a reasonable way to conduct service failover and redundancy by providing a both routing and queueing. I do think the issue is Service Management (encompassing SLA's, monitoring, alerting, failover, etc.). XI is where I think it could be done most consistently, but we are not all the way there. I will do some digging and get back to you on this with an update.

Each of these tasks could be done as Infrastructure Services supported out of WebAS 6.40. This might not be a bad move for the next 6-12 months. If these services are modeled after the standards, then a simple form of SLA monitoring, logging, and alerting should be pretty easy to create and maintain. It is pretty straight forward to say that all your service calls go through a set of basic infrastructure services to provide the visability you are requesting.

Excellent question, that deserves a better answer. I will see what I can find out.

-- David

[Former Member]

Hi David,

...It is never any fun to say "It depends"." - fully agree

As the management level is probably the one who needs to get convinced that ESA is the right way to go I am very hesitant to use answers like "we build something on top of WAS".

What I am looking for is an answer which addresses "How can I implement ESA - dare to say it - "out-of-the-box" using the outside-in approach.

Architecture without a blueprint for implementation - nothing you could sell to customers.

I am excited to hear about your findings..

Regards Rudi