Real-Time Application Maps for Proactive and Actionable Visibility
summary
Claims a working pipeline that builds real-time application maps by merging Kubernetes config, Prometheus metrics, cloud APIs, traces and network flows into a graph so teams see interservice dependencies and control inputs. Implementation uses open collectors behind existing monitoring (Prometheus, kube APIs, cloud vendor APIs), a time-series store plus graph models, and rule-based + ML inference to reduce an NP-hard dimensionality problem; one component example has 160+ metrics. Product runs with customers in early production and surfaces deviations for causal analysis.
AI-written summary · about
chapters (14)
- 0:00Intro
- 0:43Challenges for Enterprise Ops
- 1:46State of Visibility and observability
- 2:27What We Are Told Would Work
- 4:47Observability and Controllability
- 7:04Data to Address the Gap
- 8:25A Way to Get There: Application Maps
- 12:16App Map: 3-Layer Map
- 12:49App Map: Flow 2
- 13:34The Use of ML + Expertise
- 15:00Recommendations
- 20:57App Map: Application Layer
- 22:11Building Application Maps
- 24:59App Map: Component Interactions
related talks
24mPractical Kubernetes Monitoring with Prometheus
56mFull-stack observability and application monitoring with
31mReal-Time Topology to Connect Platform Engineering & SRE
20mSuper vision smart goggles
53mFull-stack observability in a world that prioritizes applications (PRT201)
41mMetrics Applied
24mObservability in the MLOps Lifecycle with Prometheus
31mVisibility from Code to Cloud - How pipeline analytics changes how we think about SDLC