As part of the time-series schema, min time is embedded into _id. We can add the max bucket range to this to get an overestimate for max time. In addition to using the control.max.time field, we should map max time predicates to predicates on time to the min time portion of _id plus the bucket range. The 'plus' can be realized by subtracting from the user-provided predicate value in the query but we should guard against underflow. This should lead to guaranteed index utilization. This should be done inside $unpackBucket's doOptimizeAt() for $eq, $gt, $lt, $gte, $lte only.
- related to
-
SERVER-54398 Extend query planner to generate bounded collection scan plans for time series collections
-
- Closed
-
-
SERVER-55576 Optimize queries on time-series collections which request the most recent value
-
- Closed
-