Implement a correctness-safe Bloom filter lookup with Amazon ElastiCache for Valkey and Amazon Aurora PostgreSQL

A three-tier architecture combines a Bloom filter, an exact-match cache, and a relational source of truth to deliver correctness-safe membership lookups (e.g., 'is this card blocked for this merchant?') with sub-millisecond P50 latency. Amazon ElastiCache for Valkey hosts both the Bloom filter (BF.EXISTS/BF.ADD) and the exact-match cache on one cluster, while Amazon Aurora PostgreSQL serves as the authoritative store. The design covers composite key encoding for scoped queries, bucketed Bloom filters to avoid per-scope filter proliferation, write-path ordering to prevent stale negatives, TTL-bound cache staleness, a recent-writes safety net to close propagation lag, and rebuild isolation on Aurora read replicas. At a 1% false-positive rate and 95% negative-lookup workload, the Bloom filter absorbs over 94% of traffic and reduces database load by about 99%.

Source: https://aws.amazon.com/blogs/database/implement-a-correctness-safe-bloom-filter-lookup-with-amazon-elasticache-for-valkey-and-amazon-aurora-postgresql. 8 Sync News only summarizes and links out; content copyright belongs to the authors and original sources.