I have a flow that runs slowly based on the Apply to Each that I have built. The loop is built with only 4 actions, runs concurrently with 2 degrees of parallelism, and is mirrored with 3 other parallel branches for a total of 4 Apply to Each actions running simultaneously (2 degrees of parallelism each). I can't increase the degrees of parallelism as I run into SharePoint API call limitations very quickly with anything higher. They run in variable loop length based on outside input, up to the max of 1250 iteration. I initially built the flow with a single Apply to Each and other actions were based on it with the 5000 max. I segmented the Apply to Each to cover all 5000 (max) for easier error handling in the future.
When the flow runtime is long (many hours, close to, or even over an entire day) I expand the branches (usually 2, sometimes 3) to look through each iteration to find at least one example of what is causing them to run slowly; however, each loop count I look at has very short run times, with each of the 4 actions taking only 0-4 seconds on average, and it's impractical to look at thousands of iterations to find one example that causes the slow performance.
Is there a way to search for the outliers in an efficient way to troubleshoot better? A run export or something else, maybe?
