Array sorting and percentiles

The need to sort arrays and calculate summary data, such as the 99th percentile, is common. Recent versions of the MongoDB aggregation framework offer enhanced capabilities in this area. This example will guide you through implementing this using both an earlier and a more recent version of MongoDB.

Scenario

You've conducted performance testing of an application with the results of each test run captured in a database. Each record contains a set of response times for the test run. You want to analyze the data from multiple runs to identify the slowest ones. You calculate the median (50th percentile) and 90th percentile response times for each test run and only keep results where the 90th percentile response time is greater than 100 milliseconds.

Note

For MongoDB version 5.0 and earlier, the example will use a macro function for inline sorting of arrays. Adopting this approach avoids the need for you to use the combination of the $unwind...