This content has been marked as final. Show 5 replies
For performance tuning you will need to look at a lot of other factors:
What is the payload size? (average and maximum)
What is the configuration of OSB domain? (Cluster, number of machines, number of managed servers, Heap size of each managed server, CPUs on the machines etc)
What is the transport of Proxy and Business service?
What are the configurations of services? (Caching for ex. )
What exactly you are doing in the message flow in the proxy?
And most important of them all, how are you testing the external service and the Proxy on ALSB?
Information about the system is as follows
1. Payload request doent exceeds 1MB
2. Transport protocol used for the Proxy and business Service is HTTP
3. No Special Security machism added for the test of the service.
4. We use our own Envelope Structure containg the information about the request and service to be called.
5. Branching Technique is used to segrigate the message and call appropriate business service.
6. Proxy service just accepts the request and with in the pipeline pair we just assign the request in time and call the appropriate business service.
After the call response receive we just note the response time for each request.
By Sending Multiple request using SOAP UI Load Test, We test the service. After 10 to 45 mins of total load on the server we calculate based on numbers of request with the timetaken in seconds.
We use Weblogic 9.* . We Used default domain creation with Clustered mode and thread size to be 5 MB.
Thanks & Regards,
Can you also please post the details in the Monitoring Dashboard of ALSB when you run the test. (Enable the monitoring on Proxy and Business service if not already enabled).
I need details from Dashboard>Service Health for your Proxy and Business services, specifically the average response time.
Also, what is the strategy you are using in the Load Test of SOAPUI?
If you are using Simple strategy then definitely TPS will be lower. Reason is the additional time taken on the ALSB Layer. In simple strategy a thread sends a new request only when previous request is complete. Since you are adding an additional network hop and additional processing by adding ALSB layer, the overall processing time as seen from client(SOAP UI) will be more and less number of requests can be send by a thread in the same time as compared to directly calling the legacy service.
Said that, if the processing time on ALSB is too high then one needs to see what is taking long time and see if it can be reduced. Monitoring stats mentioned above will help in identifying this.
If you want to see how much ALSB CAN handle then you need to use Fixed Rate strategy in Load test. This way SOAP UI will fire same number of requests irrespective of the completion of previously sent requests. To check the limit of load your domain can handle what you will need to see is if the ALSB domain gets any performance problems because of that and if requests start to fail.
What action are you using to call the Business Service? Route node or Service Callout? Route node will be better in terms of performance.
Thanks for the support and sorry for the delayed in response.
For Increasing the TPS we have changed the Deployment strategy.
Details for the above are
We use Service Callout for the Business Service.
I will check with RouteNode in the next implementation and get back to you.
Please do also suggest how you will identify the root cause by seeing the Dashboards. Any Articles On Performance Tuning and Design patterns of ESB would be benificial.
So please do guide.
Thanks & Regards,