Distributed tracing has seemingly become the biggest hook for adoption of observability tools. Trace waterfall renderings provide viscerally visual insight into the inner workings of our applications. However, they're relatively expensive to ingest, process and store—especially at scale. Managing the sheer volume of spans generated at scale can lead us to discovering complex dynamic sampling techniques; which can really mess with our mental models.
Another strategy exists: the single wide event. Using only a single event for each valuable business transaction, and packing it with fields, we can reduce event volume at scale so you don't have to sample them. In this talk, I'll cover how my team made use of this technique at CircleCI to great effect, with real examples of questions this approach has allowed us to answer.
Please register for o11ycon+hnycon first, then register for this workshop. Conference registration is required.
Michael is a Platform Engineer at Honeycomb.io. Has worked with various public and private cloud providers over for the past 8 years. Originally was deeply rooted in system administration but has since gained fondness for infrastructure as code and developer tooling. He has been using Kubernetes + Terraform software pairing since 2017. In his spare time he is an avid PC gamer, enjoys cooking and tinkers with mixed reality.
We use cookies or similar technologies to personalize your online experience and tailor marketing to you. Many of our product features require cookies to function properly.