How do you handle historical data that is infrequently accessed?

For historical data that is rarely accessed, the goal is to balance cost, durability, and retrieval latency. First, classify the data by access patterns and retention requirements. Then, use tiered storage: keep active data in high-performance tiers (e.g., S3 Standard or SSD-based storage) and move older data to cheaper, low-access tiers such as S3 Standard‑IA, One Zone‑IA, or S3 Glacier. Implement lifecycle policies to automate the transition between tiers based on age or usage metrics. Store metadata in a catalog (e.g., AWS Glue) so queries can locate data without scanning the entire dataset. For analytics, use a query engine that supports partition pruning and can read from multiple tiers, like Amazon Athena or Presto. If the data must be retained for compliance, ensure the chosen tier meets durability and availability SLAs. Periodically review the lifecycle rules to adjust for changing access patterns and cost structures.