Choosing Structures Under Constraints

Selecting between a Bloom filter and a skip list based on real system constraints like memory, deletions, and ordered queries.

You are designing a database's write-path component that must (1) answer 'might this key exist?' with zero false negatives to skip disk reads, (2) tolerate a tunable false-positive rate to save RAM, and (3) support ordered range scans over recently written keys held in memory. Which combination best fits these constraints?