● Covered by 1 source · 1 reportMedium impact1 neutral
Atlassian Migrates Metrics Pipeline to OpenTelemetry for 100,000 Hosts
Key points
- Atlassian migrated its metrics pipeline from gostatsd to OpenTelemetry.
- The migration covered 100,000 hosts across 14 regions.
- The new pipeline supports traces and logs, unlike the old UDP-only system.
- CPU cost savings of 3.9% were achieved for expensive microservices.
Metrics Pipeline Modernization
Atlassian has documented its transition from an internal gostatsd-based metrics platform to one built around OpenTelemetry. This change affects a system that processes data from approximately 100,000 hosts across 14 geographical regions. The existing service maintained a 99.95% Service Level Objective (SLO), requiring the migration to occur without disrupting production alerts.
Addressing Limitations and Adopting OpenTelemetry
The previous gostatsd system, while reliable, was limited to UDP and could not handle traces or logs. With more services generating OpenTelemetry data, continuing with gostatsd would have necessitated duplicating features already under development by the OpenTelemetry Collector community. The migration aimed to overcome these limitations and align with industry standards.
Strategic Migration Approach
To minimize disruption, Atlassian avoided an immediate, organization-wide client migration. Instead, it maintained the existing StatsD-over-UDP interface and rebuilt the pipeline behind it. This approach shifted the migration responsibility primarily to the platform team, rather than requiring thousands of services to update their StatsD clients to OpenTelemetry SDKs.
Modular Collector Distributions
Atlassian implemented purpose-built OpenTelemetry Collector distributions for distinct stages: collection, ingest, aggregation, and forwarding. This modular design, mirroring the Collector's pipeline model, allows for independent modifications to different parts of the platform. This structure enabled engineers to update one stage without impacting others.
Collection Tier Improvements and Cost Savings
In the collection tier, Atlassian replaced the gostatsd sidecar with an OpenTelemetry Collector distribution already in use by its tracing team. Applications continued sending StatsD packets to the same address, while an OTLP receiver accommodated newer services sending OpenTelemetry metrics. Integrating metrics into the tracing sidecar resulted in an average 3.9% CPU reduction for Atlassian's most resource-intensive microservices, leading to an estimated 30% sidecar cost reduction across the fleet. An OpenTelemetry Lambda extension provides the same interface for serverless workloads where sidecars are not feasible.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
The daily brief
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Today's brief
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's brief
New every morning, and the back catalogue is archived by date.
Primary sources
GitHub atlassian-labs/atlassian-aggregation-processorReporting from
Atlassian rebuilt its metrics pipeline, moving from gostatsd to OpenTelemetry, across approximately 100,000 hosts in 14 regions. This migration allowed the company to support traces and logs, integrate with OpenTelemetry data, and reduce CPU costs by 3.9% for key services.