HRL must store full historical telemetry, train on the previous season only, and plan for growing volume and schema. Which storage solution fits best?

BigQuery is ideal for this scenario. It's a fully managed, petabyte-scale data warehouse designed for analytics, handling large volumes of historical data efficiently. Partitioning by season optimizes query performance and cost for training on specific seasons. Its flexible schema allows for schema extensions as data evolves. Firestore is a NoSQL document database, not optimized for large-scale analytical queries across historical data. Cloud Spanner is a globally distributed relational database, overkill and more expensive for purely analytical historical storage. Cloud SQL is a relational database not designed for petabyte-scale analytics and managing separate instances per season would be operationally complex.