The dynamic time-series bucket calculation, from getCacheDerivedMaxSize and introduced in 6.2 (SERVER-69899), decides to artificially lower the maximum size of buckets based on two relatively unchangeable characteristics - WT cache size & bucket cardinality.

For some workloads, the pre-configured slope between limits may not be desirable, as it assumes keeping all buckets in cache will suit their needs, and combines the global cardinality of buckets across all collections. The dynamic bucketing can be limited by increasing the value of timeseriesBucketMinSize as high as timeseriesBucketMaxSize (which effectively turns dynamic bucketing off).

The proposal here is to add a server parameter to influence or directly control this calculation so that a customer can adjust the sliding window between bounds to fit their scenario better. Allowing the constant factor of 2 to be adjustable OR simply adding a constant (defaulted to 1) to the front would allow a customer to adjust the slope of the calculation between min and max.