Preventing Cache Penetration in Spring Boot Using Redis and Bloom Filters

Preventing Cache Penetration in Spring Boot Using Redis and Bloom Filters Cache penetration occurs when high-frequency requests query non-existent keys, bypassing the Redis cache completely and hitting the relational database directly. Here is how we set up a Bloom Filter guard layer in front of Redis and PostgreSQL. 1. The Bloom Filter Guard Concept A Bloom Filter is a space-efficient probabilistic data structure that tests whether an element is definitely NOT in a set or MIGHT be in a set. @Component public class CachePenetrationGuard { private final BloomFilter < String > accountFilter ; public CachePenetrationGuard () { // Expected insertions: 500,000, False positive probability: 0.01 (1%) this . accountFilter = BloomFilter . create ( Funnels . stringFunnel ( StandardCharsets . UTF_8 ), 500000 , 0.01 ); } public void registerKey ( String accountId ) { accountFilter . put ( accountId ); } public boolean mightContain ( String accountId ) { return accountFilter . mightContain ( accountId ); } } 2. Service Layer Verification Before querying Redis or PostgreSQL, verify with the Bloom Filter: @Service public class AccountService { private final CachePenetrationGuard guard ; private final RedisTemplate < String , AccountDto > redisTemplate ; private final AccountRepository repository ; public AccountDto getAccount ( String accountId ) { // Step 1: Bloom filter pre-check if (! guard . mightContain ( accountId )) { return null ; // Instant rejection, saves DB from unnecessary lookups } // Step 2: Redis lookup AccountDto cached = redisTemplate . opsForValue (). get ( "acc:" + accountId ); if ( cached != null ) return cached ; // Step 3: DB fetch and cache populate AccountDto dbResult = repository . findByAccountId ( accountId ); if ( dbResult != null ) { redisTemplate . opsForValue (). set ( "acc:" + accountId , dbResult , Duration . ofMinutes ( 30 )); } return dbResult ; } } 3. Summary Combining Bloom Filters with TTL jitter in Redis shields backend databases from cache