SESSION

Using Flink and Queryable State to Buffer High Frequency Time Series Data

Flink's streaming API can be used to construct a scalable, fault tolerant framework for buffering high frequency time series data, with the goal being to output larger, immutable blocks of data. As the data is being buffered into larger blocks, Flink's queryable state feature can be used to service requests for data still in the "buffering" state. The high frequency time series data set in this example is electro cardiogram data (EKG) that is buffered from a sample rate in millisecond into multi minute blocks.