This error aside, the documentation is excellent: detailed and filled with examples and illustrations which do a good job of explaining some complicated topics.
Once installed and registered, the product obtains new spam signatures which are automatically updated every 15 minutes.
By adjusting configuration options in Brightmail's configuration file, the server can be set to tag spam, add headers, delete it, forward it to another account or save it into a disk spool for offline processing. By default, junk mail is tagged with "" in the subject line.
Most configuration is done on the command line through text tools. Most default options make perfect sense. Keeping track of complex modifications would be difficult, but most customers are unlikely to need to stray far from the out-of-the-box settings.
Custom rules can be created using Sieve scripts, which can be used to create whatever special cases your environment requires.
This is pretty cumbersome however, and although powerful, we missed the intuitive tools offered by other products.
Similarly, although the reporting agent is able to parse the logs and extract useful data for analysis, it presents only raw data, requiring offline modeling for correlation or presentation purposes.
In testing, the server's performance took a noticeable hit in delivery mail, with the filtering processes taking their toll. It never became unresponsive though or caused mail to back up. Scanning was excellent: the up-to-date filters caught the vast majority of recent spam and there were no false positives at all.