Grafana Mimir 3.0 has launched, bringing significant upgrades to its open-source, multi-tenant time series database. Designed for long-term storage of Prometheus and OpenTelemetry metrics, Mimir has become a popular choice, boasting over 4.7k GitHub stars and 30 project maintainers. The latest version focuses on enhancing reliability, performance, and cost efficiency, allowing users to scale to 1 billion active series or more. Key improvements include a new query engine that streams results instead of loading entire datasets into memory, which boosts execution speed and reduces resource usage by up to 15% in large clusters. The architecture has also changed significantly: read and write operations are now decoupled using Apache Kafka as an asynchronous buffer. This adjustment allows for independent scaling, meaning heavy query loads won’t disrupt data ingestion, resulting in smoother operations for large deployments. The Mimir Query Engine, introduced in version 2.17, is now the default engine in 3.0. It processes queries in a streaming manner, cutting peak memory usage by up to 92% compared to the previous PromQL engine. Users planning to upgrade should prepare for these architectural changes and ensure they have the necessary infrastructure in place, including access to both old and new Mimir clusters. Overall, Mimir 3.0 positions itself as a more efficient and reliable solution for handling time series data.