Recap: How Peloton Cycle does SRE
Last week, Loggly presented at a webinar hosted by Amazon Web Services and featuring Bryan Tinsley, site reliability engineer at Peloton Cycle. We had a great experience working with Bryan and the team at Amazon, so I wanted to share some of the key learnings.
You can watch the webinar here:
1. Observability is king.
With a service that streams video to thousands of concurrent users and processes nine personal metrics for each rider, updated every second, Peloton knows a thing or two about complex data processing. The mission of its site reliability engineering team is to provide tooling, knowledge, and insight that allow engineers to run their applications in production more easily and reliably.
Peloton recognizes the value of log data for observability. The company logs data coming from individual bikes and a host of other things that could affect a rider’s experience. It logs requests and responses for every third-party API call. The company has started working on client-facing metrics on the user experience. And all of this data gets sent to Loggly.
Peloton learned that being able to observe how their systems run in production gives them a leg up on multiple key performance indicators. The engineering team can deliver a better experience to their members, they can innovate faster, and they can solve problems faster, shortening mean time to resolution.
2. Log monitoring works best when it’s integrated with the DevOps toolset.
The Peloton team uses Loggly side-by-side with other mission-critical solutions like Datadog, PagerDuty, and Slack. For example, log insights help the team look more deeply at metrics tracked in Datadog. While metrics do a great job of showing the big picture, they may or may not accurately represent the experience of an individual user. Alerts get routed to PagerDuty and/ or to Slack channels so that the right people get engaged immediately when something looks out of the ordinary.
3. More logging means more innovation.
With comprehensive logging and powerful yet easy log analysis from Loggly, Peloton is able to get code into production faster. The engineering team has the insights it needs to do solid stress testing, the confidence to put code into production knowing that it will be able to immediately detect issues, and the ability to test until it has the best feature set. According to Bryan Tinsley, the more you log, the more code you can push out, knowing that you’ll be able to address any problems if they crop up when the code is in production.
4. Log management is also a big data problem.
Manoj Chaudhary, our CTO and vice president of engineering, did a deep dive into how Loggly works during the webinar. If you’re curious about the underpinnings of our service, I encourage you to watch.
You’ll also get a handy overview of the DevOps services and best practices that Amazon Web Services offers.
Give us feedback!
Would you like to see more webinars like this from Loggly? Let us know in the comments.