[documentation:anti_spam:rbls:all:recommendations]]
Dice lo siguiente.
At the MTA level
As soon as the originating mail server is listed in the chosen RBLS, the mail is rejected. This means that you will never see this mail. The risk of false positives are high, unless you check your logs very carefully. Many organizations use spamhaus in their MTA.
At MailScanner level
When the originating mail server is listed in the chosen RBLS, it is identified as spam by MailScanner and the action is customizeable. The risk of false positives are also high, but you can, for example, just tag the messages and still deliver, unless it is listed in more than x lists.
At SpamAssassin level
When the originating mail server is listed in the chosen RBLS (there are many of them activated in the default setup), it only increases the score by a determined amount, depending on the effectiveness of the list. This method is usually the most accurate. However, note that from the MTA level to SpamAssassin level, the ressource usage is increasing, so the processing time used with SpamAssassin-based RBLS check.
To summarize
When using RBL checks at MTA level, you save a lot of resources, but you must very carefully select the RBL you use. When using RBL checks at MailScanner level, you have more control on what happens to the messages identified as spam by RBL checks. Here again you must carefully choose the lists, but since you have more control over what happens to the messages, it is less critical than at MTA-level. The counterpart is more ressources needed for the treatment of messages. Finally, SpamAssassin usually gives the more accurate and up-to-date treatment, but consumes the most resources.
