Use the data block selectivity criterion (the percentage of the table the request accesses) to give more resources to requests that access a small subset of the table. For example, you can send a request doing a full-table scan against a large table to a low-priority workload, and send a request accessing only one partition (1/200th of the table) to a high-priority workload.
Data block selectivity is particularly useful on large partitioned tables, but it is also useful for nonpartitioned tables.