At Tesla we built a high throughout, highly efficient timeseries database on top of HBase. In this talk we'll describe what makes it capable of ingesting up to a trillion timeseries records per day. We'll describe the design decisions that allows for spaceefficient storage of records with rich metadata. Finally, we'll cover design choices that boosted the adoption of the timeseries store across the organization and allowed us to measure and maintain high quality of the service in a multitenant environment.