Slice queries can't optimize based on timestamp except for rare cases (
CASSANDRA-4116). However, many common queries involve an implicit time component, where the application author knows that he is only interested in data more recent than X, or older than Y.
We could use the per-sstable max and min timestamps we track to avoid touching cold data if we could pass a hint to Cassandra about the time range we care about.
|Always promote row tombstones||Resolved|