More Performance feature offerings from PI 7.3

[Former Member]

Apart from the single ,dual stack installations,IDOC_AAE ,HTTP_AAE adapter features there are lot of other interesting features in store.

By this time you all must have seen lot of blogs speaking about the new offerings of the PI 7.3,but there are lot more promising stuff towards stability and performance.

In this blog I want to speak about the new runtime parameter configuration options

1. Error Queue

2.Queue Balancing.

Are you monitoring the tRFC queue very often to know if any failure of messages in the queue blocking the other messages ? you can relax now with the new Error Queue feature of the 7.3.

Error Messages will be moved to the Error Queue from the regular queue so that other messages which are in queue need not wait till that error message is cleared.

What about those error messages ? we can set those messages either retry automatically or manual intervention for reprocessing.

All we need to do is to set the parameter EO_RETRY_AUTOMATIC to 0 or 1.

0 - Manually need to restart, 1- Automatic restart.

if you set the parameter value as 0 , in this case also you don't need to worry about the pile of error messages there is one more parameter EO_RETRY_AUT_COUNT. Using which we can set the limit the no of messages after which automatic retry has to be activated.

Queue Balancing.

Sometimes due to large message processing other messages in the queue need to wait till the larger message gets processed. As a result waiting time gets increases and performance/throughtput gets decreases.

Load balancing of queues can be acheived from PI 7.3 onwards with the configuration of the below parameters.

EO_QUEUE_BALANCING_READ and EO_QUEUE_BALANCING_SELECT.

with the configuration of above parameters messages in the longer queues will be assigned to the shorter queues ,thus results in load balancing of the queues and can improve the performance.

The above tuning parameters needs to configure in the integration engine runtime category.

This applies only for the dual stack installations as the queues mentioned above will be part of the pipeline steps of ABAP Stack.

Need still more info on this and eager to implement in your project then need to wait for the release.