OpenTelemetry and Prometheus are getting along. What’s still missing?
The New Stack Bill Doerrfeld
OpenTelemetry and Prometheus are finally playing nicer together. That helps, but teams still want cleaner data models and fewer naming headaches.
Based on reporting by The New Stack, Bill Doerrfeld — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenTelemetry and Prometheus have lived side by side for a while, but not always comfortably. A new 2026 survey suggests that’s changing. Nearly half of respondents now mix the two styles for infrastructure metrics, and 30.7% do the same for application metrics. The average ease-of-use score for running them together climbed from 3.1 to 3.6, while the share of people calling the combo hard to use dropped from 29% to 10%.
That sounds like boring plumbing, and in observability boring is progress. Contributors Dhruv Ahuja of SigNoz, Andrej Kiripolsky and Arthur Sens of Grafana Labs, and Ana Muenz say two years of interoperability work is paying off. Still, the survey makes clear that the job is not done. Users want the projects to line up better on data models, handle resource attributes and metadata more cleanly, and stop tripping over naming and formatting differences.
Atlassian’s migration shows why that matters. The company moved its metrics pipeline from gostatsd, its open-source Go take on StatsD, to OpenTelemetry while processing data from about 100,000 hosts across 14 regions. The key trick was to change the collection and pipeline machinery underneath without forcing service owners to rewrite how they sent metrics. Atlassian calls that a platform-team migration, and the payoff was concrete: aggregation now uses about half the CPU for the same traffic, sidecar costs are down roughly 30% at fleet scale, and ingest shards share CPU more evenly.
New Relic’s 2026 Observability Forecast adds another reason teams are leaning in. Based on a survey of 2,575 IT and engineering leaders and practitioners, it found that 73% are standardized on OTel, migrating to it, or testing it. The report also ties observability to AI work: 83% say it is essential for AI-generated code, and organizations monitoring AI agents are twice as likely to report a threefold return on observability investment as those that don’t. That is a very cloud-native way of saying the bill for ignoring your systems keeps getting larger.
My take — AI-written commentary, not fact-checked reporting
This is the part of open source that actually matters: not heroic rewrites, but making two standards cooperate without turning every team into a special case. The industry loves to declare victory once a project graduates; the harder work is getting the data models, metadata, and naming to stop fighting in production. That’s where real adoption lives, not in the keynote slides.
Read more about this at: The New Stack