To enhance performance for a large table with historical data that isn't frequently accessed, what should you do?

Partitioning a large table is a highly effective way to enhance performance, especially when dealing with historical data that is not frequently accessed. By partitioning the table, you effectively divide it into smaller, more manageable segments based on certain criteria, such as date ranges or categories. This allows the database to access and process only the relevant partitions during query operations, which can significantly reduce I/O overhead and improve query response times.

In scenarios where the table contains vast amounts of historical records, partitioning helps to avoid performance degradation that can occur when the database engine scans through large datasets. Additionally, when combined with indexing strategies, it can lead to even better performance for specific queries.

While other options such as compressing the table can help with storage efficiency, it does not directly enhance query performance. Similarly, rewriting queries may improve efficiency but is dependent on how well the queries are structured and does not address the inherent size and complexity of a large table. Dropping old records, although it can reduce the size of the table, may not be feasible in scenarios where historical data needs to be retained for compliance or auditing purposes. Therefore, partitioning is the optimal choice for performance enhancement in this context.