Rspamd

Rspamd is a high-performance spam filtering system designed as a modern replacement for SpamAssassin. Written in C with Lua for rule scripting, it is significantly faster and more memory-efficient than SpamAssassin while providing more advanced detection capabilities.

The filtering engine combines traditional rule-based scoring with machine learning. The neural network module learns from message classifications to improve detection accuracy over time. Fuzzy hashing identifies spam campaigns even when message content is slightly modified between sends. DKIM signing and verification, SPF checking, DMARC policy enforcement, and ARC (Authenticated Received Chain) are all built in.

Rspamd uses Redis as its backend for Bayesian statistics, fuzzy hash storage, rate limiting counters, and greylisting state. This architecture is much faster than SpamAssassin's file-based storage and enables sharing of learned data across multiple mail servers in a cluster.

The web-based management interface provides real-time statistics, message scanning details, and configuration management. Administrators can view spam detection rates, adjust module weights, manage whitelists and blacklists, and monitor throughput. The interface also allows scanning individual messages for testing and debugging.

Rspamd integrates with Postfix as a milter, with Exim via local_scan, and with Haraka and other MTAs. On cPanel servers, Rspamd can replace SpamAssassin using community-developed installation scripts. Plesk offers Rspamd as an alternative to SpamAssassin through its extension system.

For hosting providers processing large volumes of email, Rspamd's performance advantage is significant. It can process thousands of messages per second on modest hardware, compared to SpamAssassin's hundreds. The neural network learning and fuzzy hashing provide better catch rates for modern spam campaigns.