Mastering Apache Pulsar

Chapter 12. Operating Pulsar

As you saw in Chapter 11, deploying Pulsar can take some effort, but how can we ensure its reliability once it’s deployed and in use? First, we need to be able to measure aspects of our system. Second, we need to understand the metrics in isolation. Third, we need to understand the interaction between metrics. And finally, we need to be able to understand when the preceding points deviate from our expectations. If we take a step back to consider everything required to operate a system successfully, it may be overwhelming to us. Fortunately, Pulsar makes metrics collection simple by providing a slew of metrics by default. This chapter will walk through those metrics and provide some context on which metrics are helpful to keep an eye on if you’re new to operating Apache Pulsar.

Before we explore each section, it’s essential to understand the types of metrics provided by Pulsar:

Counter

A counter is a cumulative metric, and it increases monotonically. Counters help visualize the total number of a metric occurring over a given duration.

Gauge

A gauge represents a single numeric value. Gauges are suitable for representing what is going on at the exact moment versus a historical look.

Histogram

A histogram is a sampled representation of observations. A histogram is suitable for getting counts of metrics in a specific bucket of time.

Summary

A summary is a histogram over a sliding window.

Apache BookKeeper Metrics

Apache BookKeeper is the storage engine ...