Accelerating Thanos at Scale: Faster and Cheaper Queries With Parquet - Giedrius Statkevičius

This talk, presented by Giedrius Statkevičius from Vinted, focuses on enhancing the performance of Thanos by utilizing a Parquet-backed storage system. The speaker discusses how this approach can significantly reduce query times for large PromQL queries, achieving speed improvements of up to 10 times while also optimizing storage compression. He provides an overview of Thanos' architecture, introduces Parquet as an effective columnar format, and details its benefits for metrics workloads. The session incorporates real-life data comparisons on query latency, CPU usage, and storage efficiency, culminating in a roadmap for integrating Parquet support within Thanos and opportunities for audience involvement in future developments.