Code Red Newsletter #11

Hi there,

Greetings from Tokyo 🇯🇵

Konnichiwa from the land of vending machines, matcha-flavored everything, and mysteriously perfect train schedules.

This week I’ve been on the ground in Tokyo, Japan with part of the Dash0 team for the first-ever KubeCon+CloudNativeCon Japan - and it did not disappoint.

The conference sold out at 1,500 attendees, delivering two days of packed keynotes, hallway banter, and booth visits powered by canned coffee (of which there are, no joke, 17 varieties).

If you dropped by my talk on debugging OpenTelemetry: thank you. We unpacked all the quirks across SDKs and showed some real-world failure “fun”. The demo repository is open source, so feel free to break it at home.

Without further ado, let’s dive into the good stuff.

In Focus: Measuring What Matters

This issue zooms in on a classic observability challenge: how do you know if your telemetry is actually useful?

From the newly released Instrumentation Score to frameworks for aligning telemetry with business value, it’s clear we’re moving beyond “send it all” and toward “make it meaningful.”

Whether you're debugging a broken span or pitching execs on better dashboards, this one’s for you.

Introducing the Instrumentation Score

Not all telemetry is good telemetry. Sometimes it’s too noisy. Sometimes it’s missing critical metadata like service.‌name. Sometimes it’s just… a trace without a cause.

Enter the Instrumentation Score - a new community-driven standard for evaluating OpenTelemetry data quality. Developed by OllyGarden, Dash0, and other contributors, this score gives your services a number between 10 and 100, reflecting how well you’ve instrumented your telemetry.

Think of it as a linter for observability. But one that speaks PromQL and loves semantic conventions. It helps identify missing attributes, broken spans, and context gaps, while also benchmarking your telemetry quality across teams and offering concrete guidance for improvement.

P.S. We already have a working implementation internally at Dash0. It's saving us from ourselves. If you want to give it a try, get in touch.

Understand how it works.

Want to contribute a rule or two? The spec lives here.

Curious how it all started? Read the blog post.

Introducing Dash0 MCP Server

Your AI assistant deserves better than vague vibes and a half-baked API.

Say hello to the hosted Dash0 Model Context Protocol (MCP) Server - a new way for AI agents to interact with observability data like real engineers do. It’s like letting GPT use your UI... but safely.

Here’s what it can do:

• Navigate services and investigate failed checks with full context
• Execute PromQL queries for rich metric and trace analysis
• Use tools like Triage, Query Builder, and Service Catalog - just like you would

Want to see it in action? Read the full changelog.

See the docs on how to get started.

Building a Value-Driven Observability Center of Excellence

We’ve all built dashboards no one reads. This piece is about building something better.

Andrew Mallaband outlines how to move observability from a toolstack to a business-aligned function. Think: cross-team ownership, aligned KPIs, and a shared definition of “this is good telemetry.”

And if you're building that kind of maturity across your org, the Instrumentation Score might just be your best new friend.

Read the blog post.

Code RED Podcast #27: The Hard Truths of Modern Observability

This one goes deep.

Mirko hosts a candid conversation with Andrew Mallaband and others about what’s broken - and let’s be honest, still broken - in today’s observability setups. Rising costs, lack of correlation, unclear ownership... it’s all here.

Listen to the episode.

Bonus read, on what observability can learn from ad tech.

Choice cuts

Fresh from the signal buffet - here’s what else caught our eye.

OpenTelemetry Contributor Experience Survey

Docs still need love. SDKs have quirks. And yes, semantic conventions are a wild ride. But the community is motivated, growing, and maturing. This contributor survey is a great pulse-check on OpenTelemetry’s progress - and where it still needs support.

Read the survey results.

CNCF Slack Workspace Changes Coming on June 20

No, CNCF Slack isn’t disappearing - but change is afoot.

On June 20, the Slack workspace will move to a free plan. Shorter message history. Fewer integrations. A move to Discord is being floated, but nothing’s final at the time of writing.

Read the announcement.

CNPA is Here: Certified Cloud Native Platform Engineering Associate

There’s now a certification for platform engineers! The CNCF just launched the CNPA exam, tailored for engineers working across Kubernetes, CI/CD, and observability. It’s another sign that platform engineering is no longer a trend - it’s a career path.

Explore the certification.

That’s a wrap from Tokyo - where you can debug a trace and grab a hot corn drink from a vending machine before your pod restarts.

From scoring telemetry to supercharging AI agents, the theme this week is clear: don’t just collect data. Make it count.

See you in two weeks.

Kasper, out!