Advanced Scaling and Monitoring

There are many important operational metrics to measure when assessing the performance of a running Apache Flink Application. In this section, we will look at the important CloudWatch metrics for Kinesis Data Analytics for Apache Flink applications, what they mean and what appropriate alarms might be for each.

Picture of CloudWatch Monitoring Dashboard


Next, we will utilize these metrics to influence the scaling behavior of the Apache Flink application. Using autoscaling groups, we will see how to utilize the numRecordsInPerSecond metric to scale up or down Flink Applications automatically.

CW Alarm


Finally, we will dive into the Apache Flink Dashboard to look at Backpressure, Checkpointing and other Operational Performance Indicators.

Picture of Flink Dashboard