The talk is about Loki, an open-source project developed by Grafana Labs. Loki is a time series database for log lines, allowing users to store and query logs efficiently. The speaker explains how Loki works differently compared to other systems and the advantages of its design choices. They discuss the architecture of Loki's ingestion and query paths, highlighting the minimal indexing and use of object storage. The speaker also mentions the ability to handle schema updates and data migration with zero downtime. The talk concludes with an invitation to join the Loki community.