Tasks Performance

[WiskiOne]

Hi all,

We have experienced that in our application, at specific times of high workload where many processes are executed simultaneously, some tasks in the workflow that involve loading data into the cube remain in the ‘Running’ state for hours. I wonder how to find the cause of this issue, which seems to be a performance problem during periods of high workload.

Thank you.

[pranav46]

Hello, We ran into same issue in Development env. Once we execute workflow, data management job which execute BR was going into queue. You can find out using task details. If you have admin access then you can see all users activity.

If I remember the conversation we have had onestream team that dev has only one batch server so process will go into queue. We still having issue in dev due to limited resources and trying to work with onestream if we get larger platform.

Here is the link which you can help 

Application Server Overview (onestreamcloud.com)

[SH_INT]

If you have high workloads at certain periods then depending on your Environment configuration you could certainly experience resource contention issues. If there is not enough resources available to process all tasks then some tasks will be queued. In the first instance, the task log gives some useful information such as the queued time for a task and also what server the task was assigned to and you can see how many other tasks were executing on the server at the same time. For further details you could look at the Standard System Reports provided by OneStream.  

[FredLucas]

Also, another important thing to have into account is the data loading approach.

In case you are trying to load all entities via the same WF Profile step, depending on the data volume and number of entities, that could highly impact performance as OS is not able to parallelise the load. Instead, consider setting up one load step per Entity / group of entities so these are processed in parallel.
There are ways you could still trigger all the different loads centrally (e.g.: via a dashboard button that triggers a batch load) if required.

One other point would be to review what servers are being used for those heavy tasks e.g.: if triggered via a button have the load running on the Stage server and distributing the other activities between the DM and Consolidation servers as appropriate.

[SH_INT]

Agreed, I'd inferred from the original post that the loads were hanging and not even running at times of heavy load, but data loading approach is definitely another bridge that will need to be crossed to ensure optimal performance 🙂