A presentation at PlatformCon 25 by Kasper Borg Nissen
Bridging Platform Engineering and Observability: Empowering Developers with Insights Kasper Borg Nissen, Developer Advocate at @phennex kaspernissen.xyz
Who? Developer Advocate at Dash0 KubeCon+CloudNativeCon EU/NA 24/25 Co-Chair CNCF Ambassador Golden Kubestronaut CNCG Aarhus, KCD Denmark Organizer Co-founder & Community Lead Cloud Native Nordics
tl;dr ● OpenTelemetry is standardizing telemetry collection. ● Perses is standardizing dashboards. ● Applying Platform Engineering principles transforms observability from an afterthought into a seamless, scalable, and developer-friendly experience.
Kubernetesʼ Rails moment Bryan Liles at KubeCon+CloudNativeCon North America 2019 @phennex kaspernissen.xyz
To some degree, Platform Engineering is enabling the Rails moment for Kubernetes in many organizations. @phennex kaspernissen.xyz
The Principles of Platform Engineering Self-Service Experience Explicit and Consistent APIs Paved Paths/ Golden Paths Modularity Platform as a Product Core Requirements @phennex kaspernissen.xyz
@phennex kaspernissen.xyz
Observability is still fragmented Metrics @phennex kaspernissen.xyz Logs Traces Image by pngtree.com
Observability is still fragmented We donʼt have a metrics problem, or a tracing problem. We have systems problems. Metrics @phennex kaspernissen.xyz Logs Traces Image by pngtree.com
Observability is often fragmented, leading to @phennex kaspernissen.xyz Image by pngtree.com
Observability is often fragmented, leading to Complex Query Languages @phennex kaspernissen.xyz Image by pngtree.com
Observability is often fragmented, leading to Complex Query Languages @phennex kaspernissen.xyz Vendor lock-in Image by pngtree.com
Observability is often fragmented, leading to Complex Query Languages @phennex kaspernissen.xyz Vendor lock-in Metadata Inconsistency Image by pngtree.com
Observability is often fragmented, leading to Complex Query Languages Vendor lock-in Metadata Inconsistency No instrumentation due to high complexity @phennex kaspernissen.xyz Image by pngtree.com
Observability is often fragmented, leading to Complex Query Languages Vendor lock-in No instrumentation due to high complexity @phennex kaspernissen.xyz Metadata Inconsistency Lack of unified insights Image by pngtree.com
We need to find the Rails moment of Observability. @phennex kaspernissen.xyz
A shift is happening. @phennex kaspernissen.xyz
A shift toward correlation Find related information @phennex kaspernissen.xyz Jump between signals Reconstruct chain of events
A shift toward standardization Instrumentation OTel API & SDK Telemetry Backends The OpenTelemetry Collector auto-instrumentation Time-series database … Log database Receive Process Analysis Tools Export Trace database Infrastructure … Kubernetes … Generate and Emit transmit Inspired by visualizations from LFS148 Collect, Convert, Process, Route, Export transmit Store & Analyze
Vendor - space Instrumentation OTel API & SDK Telemetry Backends The OpenTelemetry Collector auto-instrumentation Time-series database … Log database Receive Process Analysis Tools Export Trace database Infrastructure … Kubernetes … Generate and Emit transmit Inspired by visualizations from LFS148 Collect, Convert, Process, Route, Export transmit Store & Analyze
A shift toward… @phennex kaspernissen.xyz
1/1/20241/1/2025 Commits: 27.168 PRs+Issues: 58.508 Source: CNCF Velocity Commits: 44.486 PRs+Issues: 56.299
So, why OpenTelemetry? Instrument once, use everywhere @phennex Separate telemetry generation from analysis kaspernissen.xyz Make software observable by default Improve how we use telemetry
Thatʼs all great, but how do I make it easily accessible for my developers? @phennex kaspernissen.xyz
Platform Engineering for Observability Self-Service Experience Explicit and Consistent APIs Golden Paths Modularity Platform as a Product Core Requirements @phennex kaspernissen.xyz
Platform Engineering for Observability Self-Service Experience Auto-Instrumentation Explicit and Consistent APIs Semantic Conventions Golden Paths Observability built-in Modularity Collector Pipelines Platform as a Product Documentation + Support Core Requirements Cross-signal correlation @phennex kaspernissen.xyz
OpenTelemetry Operator Instrumentation OpenTelemetryCollector OpAMPBridge OpenTelemetry Operator @phennex kaspernissen.xyz TargetAllocator
Auto-Instrumentation with the OpenTelemetry Operator Instrumentation Instructs how to inject auto-instrumentation Injects instrumentation in to the pod OpenTelemetry Operator @phennex kaspernissen.xyz
Observability doesnʼt stop at instrumentation. @phennex kaspernissen.xyz
Perses An open specification for dashboards. CNCF Sandbox project @phennex kaspernissen.xyz
Dashboards as Code @phennex kaspernissen.xyz
Dashboards as Code Perses PersesDatasource perses-operator @phennex kaspernissen.xyz PersesDashboard
Demo PersesDatasource instr. todo-go PersesDashboard Perses Operator todo-java Perses Postgres MySQL Prometheus Inject eBPF-sidecar Inject Java Agent OpenTelemetry Operator Instrumentation @phennex kaspernissen.xyz OpenTelemetry Collector OpenTelemetryCollector Jaeger
Demo time 🙏 @phennex kaspernissen.xyz
Letʼs recap… PersesDatasource instr. todo-go PersesDashboard Perses Operator todo-java Perses Postgres MySQL Prometheus Inject eBPF-sidecar Inject Java Agent OpenTelemetry Operator Instrumentation @phennex kaspernissen.xyz OpenTelemetry Collector OpenTelemetryCollector Jaeger
The Real Goal: Observability as a Paved Path @phennex kaspernissen.xyz
Observability is evolving - fast. @phennex kaspernissen.xyz
OpenTelemetry is standardizing telemetry collection. @phennex kaspernissen.xyz
Perses is standardizing dashboarding. @phennex kaspernissen.xyz
Applying Platform Engineering principles can transform observability from an afterthought into a seamless, scalable, and developer-friendly experience. @phennex kaspernissen.xyz
Observability is a systems problem - not a tracing, logging, or metrics problem. @phennex kaspernissen.xyz
When we connect signals together, we empower developers to solve problems faster. @phennex kaspernissen.xyz
Thank you! Metrics Open Standards Traces Logs @phennex kaspernissen.xyz
Thank you! Metrics Open Standards Traces Logs @phennex kaspernissen.xyz
Platform engineers often face the challenge of providing developers with effective tools for managing application observability. Observability sprawl - characterized by multiple tools, disconnected dashboards, and inconsistent instrumentation - can lead to difficulties in troubleshooting, increased cognitive load, and reduced productivity. In this talk, Kasper Borg Nissen will demonstrate how platform engineering principles can help bridge this gap by creating developer-friendly observability abstractions. The session will cover how OpenTelemetry can be used for auto-instrumentation, manual instrumentation, and standardizing semantic conventions for telemetry signals. Attendees will gain actionable insights into consolidating telemetry data into a unified view, eliminating tool fatigue, improving debugging workflows, and creating a seamless observability experience for developers.
