Time-Based Sliding Windows

The previous examples all used record count as the basis for their window strategies. However, Data Pipeline also provides several built-in strategies for working with time based windows.

You can create adjacent windows that stay open for 10 minutes.

You can also mix strategies and create windows that open every hour until they've collected 100,000 records. Depending on how long it takes to collect that many records, you may end up with some overlapping windows and some sparse windows.

If you wanted to limit the hourly windows to 100,000 records or 15 minutes — whichever came first — you can combine a record strategy with a time strategy using the or() operator function.