How to observe an unikernel

Collect and see metrics

Unikernels have the distinctive feature of running with a certain degree of autonomy. It may, however, be necessary to monitor these unikernels - in particular, to know how much memory they are using, how they are utilising their vCPUs, and also to keep an eye on network activity.

In this chapter, we will look specifically at how to install various tools for monitoring unikernels. In this case, we will install:

• InfluxDB to collect metrics
• Telegraf, which allows us to bridge the gap between albatross_stats and InfluxDB
• Grafana, which allows us to visualise the metrics

Finally, we will use one of our libraries, tally, to generate slightly more specific metrics relating to our unikernels.