Honeycomb user stories are the best. Shining a light on a specific use-case plus following the breadcrumbs and steps to discovery and resolution is pure gold because everyone learns. An early Honeycomb customer, Nylas (and our neighbor in downtown San Francisco) shared this blog that highlights how the team optimizes performance and cuts DB queries by 50%. Thanks Jonny Hatch and the Nylas team.
Nylas APIs relied upon daily by users to sync billions of emails, cal events and contacts are stored in MySQL. Unacceptable lag means a less than optimal experience and so the Nylas eng team proactively watch performance to get ahead of any problem and reduce stress on customer support.
Instrumentation is your friend
Nylas added custom instrumentation for better telemetry which pays off. Knowing there’s a lag is just one step. Understanding the frequency plus the location of code where it’s stemming from is critical – otherwise you waste time searching and debugging.
Visualization helps to hone-in
Spikes in Honeycomb charts focus on where to drill down and being able to target a particular field is what Honeycomb does best. The Nylas team found the MySQL DB was reading all the history which introduced the lag and once validated, the team could move to finding a solution.
Working closely as a team is important not just for learning but because troubleshooting together is much easier, especially now when everyone is remote. Check out session on how to be remote but not alone with Honeycomb’s Michael Wilde.