Avoid processing of Duplicate Synchronous Messages

[Former Member]

Hi,

I have two synchronous messages coming with almost zero delay in it. Both the messages contains the same data which is causing a major problem in the backend.

I want to avoid the processing of duplicate message. My scenario is HTTP to SOAP (sync)...

The option that I thought of are:

- Using BPM so that messages can be sent sequentially to backend and there it can be checked for duplicate entries (client is not ready to use BPM)

- Providing Delay in Message Mapping...(I will end up with same issue as same amount of delay will be provided to both the messages may or may not but dont want to take risk)

- I cant make a look up to see if entry already exists as the difference between incoming duplicate messages are so small that it will not give the desired result.

Also thought of creating message ID in URL only but it is not possible with QOS BE.

Is there any way through which I can process the sync messages in sequential way... or something that will avoid processing

of duplicate messages?

Any thought?

Thanks,

Sunil Singh

[Senthilprakash1 · Accepted solution]

Hi Sunil,

Checking the data weather is duplicate or not (ie, its already resides in Rev application) would sure require to use of lookup or BPM.

>

> I want to avoid the processing of duplicate message. My scenario is HTTP to SOAP (sync)...

even if the data is comming with some delay between them. how will you check that is duplicate message or not unless you use lookup or BPM in it to compare two independent messages at runtime.

My idea would be! instead of trying to put a validation at PI.

check in the source system whats causing their application to generate duplicate messages in first place and send it to PI.

check whats the issue over there and ask them to rectify it or rather not to trigger duplicate data. (it would be easier to control this from source application).

PS: even if we find a way to stop duplicate messages to be processed inside PI. its unnecessary load for PI.

-->Senthilprakash.

[sunil_singh13 · Accepted solution]

Hi Senthilprakash,

Thanks for your reply...

In my case the source system is a JSP page which is storing nothing. so performing validation on front end is like storing the data in separate database and then build a logic. Also we already have a alogic built in the Backend system (SAP) but it is not capable of handling negligible time difference between duplicate data and perform the validation..

PI being the midiator between these two it is 'nice to have' the logic built in PI so that it will take care of the same for front end as well as backend.

Thanks,

Sunil Singh

[Harish · Accepted solution]

Hi Sunil,

We has similar issue with FTP and Mail channels and we have implemented the clusterSyncMode = lock advanced parameter.

Please check if it is available for HTTP adapter as well.

Regards,

Harish