OAC Dashboard Slowness After Migration

Alvaro_Soares

Hello All,

After completing the migration from OBIEE 12c to OAC, end users have been reporting some slowness when opening dashboards in the new OAC environment.

I followed the official Oracle documentation for executing this migration via the URL ( https://docs.oracle.com/en/cloud/paas/analytics-cloud/acmgb/migration-task-flow.html ), and in the section Clean Up and Reconfigure Your Semantic Model .rpd File ( https://docs.oracle.com/en/cloud/paas/analytics-cloud/acmgb/clean-and-reconfigure-your-semantic-model-rpd-file.html ), it is mentioned that the parameters DEFAULT_BULK_FETCH_ROW_COUNT and MAX_BULK_FETCH_BUFFER_SIZE for each physical layer should be configured to 1000 and 3276800, respectively.

However, the slowness is still noticeable in the dashboards. Could anyone advise where I can further analyze these performance-related aspects in more detail? Is there any monitoring module in OCI that allows me to track OAC instance utilization? I would also like to know if there are any options within OAC Administration that provide insights related to this.

Nevertheless, is there any way to monitor and optimize Dashboard Performance in Oracle Analytics Cloud (OAC) ?

Thank you in advance.

SteveF-Oracle

Hi Alvaro,

"Performance" can be such a large topic:
+ query
+ UI loading / navigation
+ Poor modeling
+ Poor dashboard design (huge)

A few points:

I followed the official Oracle documentation for executing this migration via the URL ( https://docs.oracle.com/en/cloud/paas/analytics-cloud/acmgb/migration-task-flow.html  ), and in the section 

Clean Up and Reconfigure Your Semantic Model .rpd File

 ( https://docs.oracle.com/en/cloud/paas/analytics-cloud/acmgb/clean-and-reconfigure-your-semantic-model-rpd-file.html  ), it is mentioned that the parameters 

DEFAULT_BULK_FETCH_ROW_COUNT

 and 

MAX_BULK_FETCH_BUFFER_SIZE

 for each physical layer should be configured to 

1000

 and 

3276800

, respectively.

We will need to get the documentation updated for this point, as we recommend higher values for datasources in remote data centers: 5000/10000000

REFERENCE: https://blogs.oracle.com/analytics/post/oracle-analytics-cloud-minimizing-impact-on-query-performance-due-to-network-latency

When you use the term "dashboards", can you clarify if you are referring to 'Classic' Dashboards (/analytics) or a Workbook (/dv), or both/all.
Also, what the datasources are (i.e. file-based datasets, database datasets, only subject area RPD, etc.)


This may influence the guidance and suggestions.
Some general guidance since we won't have any specifics here:

Isolate the issue - take one "dashboard" copy, remove things until it performs
Check your session logs and your query timings - or Usage Tracking
Check initialization blocks for failures, check if deferred, non-differed are executed at login and could delay things — test them
For UI, you can collect and review a HAR file
SRDC - How To Collect Browser Console Logs From Chrome, Microsoft Edge or FireFox Browsers KB171102

Just some quick off the cuff items to start.

Other comments, welcomed.