Infinite Slices of data, however small or large¶
InfinSlice enables users to process data that was ingested within a window of time. InfinSlcie represents slices of data between a start time and an end time. Stated another way, InfinSlice is the snapshot of the bucket at the end time, with all objects that present at the start time removed from the view.
Reading from any InfinSlice is as quick as reading directly from the bucket as there is no recovery or preparation step in the process. Users do not need to initiate an explicit infinslice action. There is no data copy involved as this is pure metadata operation.
InfinSlice is useful for:
- Data Drift Protection - retrain model against a new slice of data
- Continuous Training - continuously retrain model using small slices of data