UserBanBloomService

Info

@Service implementing a Redis-backed bloom filter with exact-set verification for fast user ban lookups.

Overview

Uses a two-layer approach: a bloom filter (Redis bitmaps) for fast negative checks, and a Redis Set for exact verification. This eliminates false positives while maintaining O(1) lookup performance for the common non-banned case.

Methods

Method	Description
isBanned(username)	Check if user is banned (bloom filter + exact set)
ban(username)	Add user to both bloom filter and exact set
unban(username)	Remove user from exact set only (bloom filter bits are not cleared)

Configuration Properties

Property	Default	Description
user.ban.bloom.key	user:ban:bloom	Redis key for bloom filter bitmap
user.ban.set.key	user:ban:set	Redis key for exact ban set
user.ban.bloom.bit-size	1000003	Bloom filter bit array size (minimum 1024)
user.ban.bloom.hash-count	6	Number of hash functions (minimum 2)

Hash Strategy

• Hash 1: CRC32 of normalized username
• Hash 2: Java String.hashCode() with #ban salt
• Combined: hash1 + i * hash2 + i * i (double hashing technique)

Behavior Notes

• Usernames are normalized to lowercase before hashing
• unban() only removes from the exact set; bloom filter bits remain set (standard bloom filter behavior)
• False negatives are impossible; false positives are eliminated by the exact set check

Dependencies

• userBanRedisTemplate -- Redis operations on dedicated DB 9