Q & A Home
Customizing SNF
Errors
False Positives
Functionality
Integration
Log Files
Resellers
Result Codes
Rulebase Updates
Software
Spam
Subscriptions
Trials
Version 3 Architecture
Documentation Home
Resellers
Do you have any performance information?
You can look at general filter performance metrics by examining our live performance statistics in the Live Data section on this site.
The SNF engine provides the fastest, most efficient anti-spam engine in the industry while maintaining accuracy that consistently matches or exceeds other engines.
For example, a single CPU (generic hardware) windows based server providing pop3 & smtp services can sustain rates in excess of 3500 messages per minute with occasional peaks to 4500 mpm.
In a recent real-world test on customer equipment: While running in parallel with SpamAssassin on the same *nix (ubuntu) / postfix system SNF required less than 10% of the CPU resources requried by SA while processing the same message stream, producing slightly higher capture rates, and producing fewer false positives.
As for spam capture rates, independent testing by Computerized Horizons (declude.com) consistently show that Message Sniffer out performs all other single tests and captures on-average 90-95% of spam.
Our own continuous monitoring systems show that SNF consistently captures > 98.5% of spamtrap content and frequently sustains capture rates > 99.6% for long periods (hours).
Anecdotal evidence from users testing Message Sniffer head-to-head with Spam Assassin find that spam capturing and false positive rates are generally comparable but Message Sniffer is significantly faster and ultimately requires less administration. Message Sniffer is also updated more frequently so it tends to capture newer spam more aggressively than SA.
False positive rates have been estimated (by users and other third parties) from .018% to 0.2% depending upon the level of tuning done and the breadth of the user base involved. 0.2% seems to be representative of the "out of the box" false rate for ISP class systems, while small business customers often report no false positives, or nearly no false positives after adding only a few white rules.
An overall metric of false positive rates may be indicated by data from our message processing telemetry and false positive reporting systems. While handling more than half a billion messages per day (at the time of this writing) we generally receive fewer than 20 false positive reports per day from all customers and their professional support staff.
For more information about SNF performance please see the Features section of our documentation.
