📊Atlassian Rebuilds Metrics Pipeline with OpenTelemetry, Cuts Costs
Atlassian slashes costs with OpenTelemetry
TL;DR
Atlassian rebuilt its metrics pipeline with OpenTelemetry, reducing costs by 30% and handling 4.8 billion data points per minute. The move preserves existing workflows and sets a new standard for large-scale observability.
Atlassian just rebuilt its massive metrics pipeline around OpenTelemetry, cutting costs by 30% across its fleet. The new system handles 4.8 billion data points per minute and stores 220 million, all while using half the CPU of the old setup. If you're running a large-scale observability system, this is a big deal. Atlassian's approach keeps familiar workflows intact while moving to OpenTelemetry, making it a practical upgrade path for others. The platform now receives data from 100,000 hosts across 14 regions, with a new Collector distribution for collection, ingest, aggregation, and forwarding.
Key Points
Atlassian's new metrics pipeline handles 4.8 billion data points per minute, storing 220 million.
The platform receives data from 100,000 hosts across 14 regions.
Atlassian cut sidecar costs by 30% across its fleet with OpenTelemetry.
The new system uses half as much CPU for the same traffic after developing an aggregation processor.
Atlassian introduced purpose-built Collector distributions for collection, ingest, aggregation, and forwarding.
Why It Matters
If you're running a large-scale observability system like Atlassian's, moving to OpenTelemetry can cut costs by 30% and improve performance. The new system handles 4.8 billion data points per minute with half the CPU, making it a practical upgrade for teams looking to modernize their observability stack. This approach preserves existing workflows, making it easier to transition from legacy systems.
Comments
Be the first to comment
Enjoyed this article?
Get it daily. 7am. Free. Reads in 5 minutes.