Atlas is an in-memory dimensional time series database backend for managing time series data.
The tool addresses the need to store and query high-cardinality dimensional metrics efficiently. It holds time series data in memory and organizes it by dimensions, allowing queries that filter and aggregate across multiple tagged attributes. This approach enables fast retrieval of metrics grouped by arbitrary combinations of tags, which is essential for monitoring systems that need to slice data across many dimensions like service, region, instance, or endpoint.
Atlas suits organizations running large-scale monitoring infrastructure where dimensional queries against high-cardinality data are frequent. It works well for teams already invested in Netflix's ecosystem or those building custom monitoring backends that require fine-grained control over time series storage and querying. The in-memory design makes it best suited for environments where the working set of metrics fits within available RAM and where query latency is a priority over long-term historical storage.
The project maintains steady development activity with regular issue engagement and ongoing maintenance. Pull requests receive consistent review and integration. The codebase shows active refinement of core functionality rather than major architectural shifts. Documentation is actively maintained alongside code changes. The project sustains a mailing list for community discussion and support.
