JDBC Receiver performance

[Former Member]

Hi,

Some quick questions concerning JDBC Receiver performance.

1. I know you can increase the standard max 5 threads for the adapter. But want to understand if you set this ex. to 20 is this then the max for the JDBC adapter on the complete server or the max per server-node on the server? So, as an example, you have 1 server with 3 nodes and set the max threads to 20, is it then 20 for the server in total or per server-node (so 60 threads in total then)?

2. We have a scenario in which messages arrive in PI on ad-hoc basis and then are inserted into a non-sap applications database into 1 single table. This works fine, except that before each insert-statement there seems to be another query performed first in their database which takes 5-10 times longer than the actual insert of our message. This query checks the table-scheme of the table in which we insert each time. How can this be prevented/switched off?

Kind regards and thanks in advance for any tips/solutions,

Kurt

[marksmyth · Accepted solution]

Hello Kurt,

1. In this case, in theory, each server node will be able to process 20 messages in parallel from JDBC adapters. Thus, with 3 nodes, you will be able to process 60 messages in total.

2. I'm not sure!

Regards
Mark

[AnilDandi]

Hi Kurt,

The query to pull table's metadata information could be the work of the JDBC driver used by PI.

Try running a trace on the driver (can be done by setting jdbcDriverTraceLevel in the communication channel, see SAP note 953120). If that is indeed the case, you may have to patch the JDBC driver.

[iaki_vila]

Hi Kurt,

About your second question, if you have an XI 3.0, this note could be helpful  870334 - XI 3.0 JDBC Adapter: Receiver executes unnecessary query

Regards.

[Former Member]

Hi, no, we have a PI 7.11

[Former Member]

Hello Kurt,

1. Yes, there will be 20 threads per node, which in your case would be a total of 60 available threads for that particular adapter type.

2. The query check on the table behavior sounds strange. How does your channel configuration of the JDBC receiver look like? Is the 'processing' and 'advanced' tabs default or do you have any custom settings?

Regards,

Prajeet

[Former Member]

Hi Prajeet,

In the Processing tab we have nothing special except "XML Schema Interpreter - Key Tags Mandatory' is checked and Interpretation of Empty String Value = NULL Value

In the Advanced nothing besides number of retries, isolation level (as specified by the Oracle db-guys), Auto-commit enabled and a poolwaitingtime.

Kind regards,

Kurt