The Bloom Filters Roadmap

Why trade a little certainty for a lot of memory, the family of probabilistic data structures, and exactly what a Bloom filter promises (and doesn't).

Derive the false-positive formula, compute memory and the optimal number of hashes for any target, predict saturation, and do the interview arithmetic in your head.

What a Bloom filter needs from its hashes, why production filters use double hashing from one 128-bit hash, and how to detect and fix bad distribution.

Implement a Bloom filter in Java with sizing, double hashing and tests; make it thread-safe; serialize and version it; and know when to use a battle-tested library instead.

Scalable Bloom filters for unknown sizes, Counting Bloom and Cuckoo filters for deletion, blocked and partitioned layouts for CPU speed, Quotient, Xor and Ribbon filters, and a decision guide against HashSet, HyperLogLog and Count-Min Sketch.

Why every LSM-tree database keeps a Bloom filter per SSTable, how RocksDB, Cassandra, HBase and Parquet configure them, and how CDNs use filters to avoid caching one-hit wonders.

The membership accelerator in front of a database, cache penetration defence against random-ID attacks, RedisBloom operations, local vs shared filters, and a username availability API.

Stale filters and the dangerous direction, keeping inserts in sync with the database, handling deletes, zero-downtime rebuilds with versioned filters, and building huge filters in parallel and merging them with OR.

Event-driven filter updates with an outbox, CDC and Kafka; per-shard filters that cut query fan-out; filters under consistent-hashing rebalances; and multi-region filters with failover.

Metrics that matter and how to measure the real false-positive rate, failure modes and fail-open vs fail-closed, fifteen failure-injection scenarios, security and membership inference, and complete capacity planning from 500M to 10B items.

FAANG-style designs: a Google-scale web crawler, a billion-user membership service and product catalogue, malware and global blocklists, a distributed object-existence check, an LSM storage engine, and the full Bloom + Redis + Kafka capstone with anti-patterns.

Fifty graded interview questions with model answers, from beginner definitions to staff-level design trade-offs, plus the final readiness checklist and a four-week study plan.