Multiple report activities in a proxy

2646335

Hi Team,

In one of our project we have multiple report activities to report the request and response and intermittent request responses of other service callouts. Will this impact the performance?

Instead of adding multiple report activities can we have one variable which store all the information and using a single report activity to publish a message report.

Please suggest.

Thanks.

vladodias

Hi mate,

Why don't you just turn on full execution and message trace for that particular service? That will probably give what you want, without having to use multiple report activities...

https://thecattlecrew.net/2015/12/23/oracle-soa-12c-quicktip-enable-servicebus-message-tracing-in-defaultdomain/

https://docs.oracle.com/middleware/12211/osb/administer/GUID-F6648561-8CB9-44C4-A63B-C68E23D1005E.htm#OSBAG2382

Cheers,

Vlad

2646335

Thanks Vlad for the reply.

The service is taking 4secs, if we enable message trace i'm afraid it will degrade the performance further. Please correct me if i'm wrong.

Out of the 4secs, 2-3secs is the response time of the target system. We wanted to improve the performance, so thought of adding one single message report instead of multiple message reports.

Please suggest.

Thanks.

vladodias

Message reports just push information into queues, the performance impact should be minimal... OSB is very lightweight overall as well...

A performance tuning exercise usually involves identification of the bottlenecks, finding the causes and correcting them... In your case I don't think the bottleneck would be message reports, so I'm guessing you will not get much improvement by turning them off...

I would be interested to know the results, but I believe that, at the best, you will get an improvement on the order of milliseconds...

Cheers,

Vlad

Martien van den Akker

Hi,

As Vald says, I don't think reports will decrease your performance significantly.

I'd like to add that the Reporting (at least in 11g) uses a topic within the transaction. So in case of an exception, the reporting will be rolled back. So if you need to store the request and correlated responses for track and tracing, then you might loose the ones that lead to exceptions. You should think about what the purpose of your reporting is.

But I agree with Vlad that you should investigate the actual functional working of your services, to see where the bottlenecks are.

Regards,

Martien