How XI sends the synchornize messages to the receiver via the ABAP proxy

[Former Member]

Hi experts,

I have some assumptions according to my understandings in XI, please take a look and correct me if some are wrong.

1.  the messages will not make use of the inbound queue and outbound queue in the integration engine as it's in synchronize mode, these queue are for EO/EOIO messages

2. if the sync messages don't use of the queues, does that mean the integration engine send the messages immediately once the pipeline services completed?

3. as we are using the ABAP proxies for the receiver system, the integration engine will send the messages to receiver directly via the http protocol instead of sending them to adapter engine, can we config the max connections for this http connection?

4. which will cause the pipeline step 'call_adapter' to take much long time?

Regards,

Bean

[kkram]

> 1.  the messages will not make use of the inbound queue and outbound queue in the integration engine as it's in synchronize mode, these queue are for EO/EOIO messages

Queues are used only for asynchronous messages. This is done to guarantee delivery. Synchronous messages are executed immediately on arrival based on available dialog processes.

> 2. if the sync messages don't use of the queues, does that mean the integration engine send the messages immediately once the pipeline services completed?

Yes, the messages are sent immediately. You may choose to persist all the messages (usually done in development systems) or persist only in case of errors (done in production systems). This setting is done in SXMB_ADM.

> 3. as we are using the ABAP proxies for the receiver system, the integration engine will send the messages to receiver directly via the http protocol instead of sending them to adapter engine, can we config the max connections for this http connection?

Yes, the message is sent using HTTP service on the ABAP stack. Max connection can be configured using profile parameter icm/max_threads as described in this link: http://help.sap.com/saphelp_nw70/helpdata/EN/ac/8f813a34872769e10000000a114084/frameset.htm

> 4. which will cause the pipeline step 'call_adapter' to take much long time?

In case of synchronous messages, since a response is expected, any delay in the backend system or even in the connection to the backend system will result in the longer running times in 'call_adapter' step. However, there is a ICM timeout that will error out the message in case of long running processes. The timeout is a profile parameter as well.

Hope the above helps.

Thanks

KK

[Former Member]

Hi KK,

firstly I 'd like to give my many thanks for your kind reply .

Regarding the question:

question 3:

Yes, the message is sent using HTTP service on the ABAP stack. Max connection can be configured using profile parameter icm/max_threads as described in this link: http://help.sap.com/saphelp_nw70/helpdata/EN/ac/8f813a34872769e10000000a114084/frameset.htm

do we have any rule for how to set the profile parameter ''icm/max_threads'?

question 4:

In case of synchronous messages, since a response is expected, any delay in the backend system or even in the connection to the backend system will result in the longer running times in 'call_adapter' step. However, there is a ICM timeout that will error out the message in case of long running processes. The timeout is a profile parameter as well.

do you mean that the step 'call_adapter' also includes the time spent for the business logic execution in the target backend system?

Best regards,

Bean