Collecting and graphing performance data

Logs are great, but sometimes what is really needed is a summary of system performance. A quick view of performance graphs can often tell an administrator much more than sifting through hundreds or thousands of log entries. A simple question such as "Why is this application slow?" can often require looking at performance data for the disks on the storage nodes, network utilization between each storage node and between the storage and the Docker host, memory and CPU utilization, and host disk performance. That does not even include things such as database speed and application inefficiencies. The good news is that all of this data can be collected and graphed.

Collecting data with collectd

The collectd daemon is a powerful tool that can collect data from many different sources and send the data to a number of databases. Two of those databases are Elasticsearch and InfluxDB. Using Elasticsearch means that the data is available for Kibana to build visualizations...