Grafana Mimir, an open-source time series database, introduces a scalable split-and-merge compactor capable of handling up to 1 billion active series, addressing limitations of existing Prometheus and Cortex compactors. The compactor employs features such as grouping input blocks, sharding series, and merging shards to enhance performance and manageability. These features allow for efficient data ingestion, compaction, and deduplication, ensuring fault tolerance and durability through 3x replication. By splitting series into configurable shards, Grafana Mimir overcomes Prometheus TSDB format limits, facilitating better horizontal scaling and improved query performance. The compactor's capabilities allow Mimir to manage the regular influx of data from ingesters, with its design supporting efficient deduplication and block management. The system's flexibility and compatibility with tools like Prometheus and Thanos, coupled with its integration into Grafana's ecosystem, make it a powerful tool for handling large-scale metrics efficiently.