JDBC vs JMS adapter performance

[Former Member]

Hi Guys,

We have an interface - to send 2 million records (10 - 12 fields) out of SAP and we are on PI 7.40.

I am trying to compare the pros and cons of JDBC and JMS adapters.

Does anyone of you have some bench marks on this topic and I would really appreciate it.

Thanks,

-Naveen.

[iaki_vila]

Hi Naveen,

If your idea is to do a migration and to use only one time the interface i think SAP PI is not the best solution and you should do a copy from your DB to the new one. However, if you are working really with a interface with 2 million records traffic per day, i think you can do easier with JDBC, as Hareesh said with 5 until 10 MB i have noticed a quicker performance, all depends about your PI hardware. You can do some test, if the time is between 5 to 10 seconds with a huge load, for example every 5 seconds and other asynch scenarios working, you can set it.

Hope this helps.

Regards.

[Former Member]

Hi Naveen,

I believe,JDBC would be better and faster as compared to JMS while handling high volumes.

But as Harish mentioned,we need to split the messages otherwise it might block some other messages.

[former_member184720]

You have mentioned out of SAP and you are referring to JDBC or JMS? Mostly we use proxy/IDOC/RFC/Soap to extract the data from SAP unless you connect to backend DB.

However it doesn't matter which adapter you use as long as you split the messages into chunks.(Ideal size ~ 5MB per message). Both should work fine.

Both the adapter/in fact PI may not handle if you send 2 million records in one message