Profiling is all about measuring applications dynamically at runtime to understand their CPU usage, memory usage, and other important metrics. Profiling can yield deep insights about your application’s performance at runtime right down to the line of code that allocated some memory, or created a thread.
Previously profiling was almost always performed on a “point in time” basis; continuous profiling improves upon this methodology by introducing a new dimension to your profiling results: time.
This talk shows how you can get started with continuous profiling by combining Juju and Parca, and how snaps & charms can ease the integration of such complex processes into your workflow.
Session author bios
Jon is VP Engineering, Enterprise Solutions at Canonical, leading the teams that deliver Juju and a suite of enterprise-ready operators for data platforms, machine learning operations, observability and more.
Following a degree in Computer Science, Jon’s career started at the Royal Military Academy Sandhurst. He joined the Royal Corps of Signals where he played a key role in helping the UK define high-level doctrine for cyber operations and electronic warfare.
While working on next steps beyond the military, Jon spent 3 months participating in the Consensys Tachyon accelerator for early-stage blockchain startups. He subsequently worked for Thales UK, where he designed and built a public-cloud first development platform for UK government projects, bringing modern infrastructure-as-code and immutable infrastructure practices to a wider audience inside Thales.
Linux, open-source and tech interests aside, Jon is a husband and father with a keen interest in coffee, cars and bikes!
|Level of Difficulty