APO and LiveCache Optimization

[Former Member]

Hi All,

Can anyone share some points to optimize APO and liveCache, like best practices, system configurations to improve the performance of our system?

We would appreciate your response and inputs.

Regards,

Philip Lledo

[Former Member]

Hi Philip,

There are definitely few things that you can check in your system -

Make sure for every long job execution step you have enough buffer left in your storage space. Monitor storage quality and determine how often it should be executed. Also avoid any locks for long time in SM12.

You may consider using external parallel processing profile (divide one big job into several jobs) for the job, which takes long run time. Find optimal package size by increasing value like 150, 200…you can do some experiment to find a better combination for your parallel processing profile. Like, Increase the number of parallel processing. i.e. 15, 20…and try to find optimal block size by reducing value like 250, 125, 60.. in your QA system. Also consider applying Note 1359613 - SNP Optimizer: Parallel processing is slow.

Check the current spool list in your job logs. See if you really need the entire spool or you can reduce the volume of the spool file or stop creating spool file as applicable.

Make sure your Load balancing for RFC is set properly in your APO System, analyze the current situation by taking help from your Basis team.

Check the RFC recourse utilization in every server path/instance. There should not be any scenario, where one instance is used almost in full, while there are many free work processes on other instances. The reason for this is that there is no load balancing defined in your APO system. There is a SAP note 593058 for more details to set the load balancing within the server group.

You can do some experiment in your optimizer profile (if you are using SNP/Deployment optimizer for planning) in QA system- In your optimizer profile, try to reduce the value of ‘Maximum Runtime’ and checked. You should consider the quality of result of the optimizer run here. Then decide if this is an option at all.

Also you may consider to reduce the horizon for your optimizer run, if applicable. May be you don’t need to run the optimizer for full planning horizon, if you are not really publishing for full horizon. Also consider to reduce your no of buckets in the planning horizon if possible, Like you don’t need to have all the daily bucket in the horizon, may be first month in daily bucket. You can determine planning horizon in the variant.

Remove any inconsistencies in LC every day after executing the report and make sure to resolve all the inconsistencies before executing major planning job. Also Monitor post processing record regularly and process them accordingly.

By checking/doing all the above, you should get some pointers.

Thanks,

Satyajit

Message was edited by: Satyajit Patra