Spam continues to be an annoyance for everyone, wasting bandwidth, processing time and cluttering inboxes; the current estimate for daily spam volume is 93Bn messages per day.
As long as there’s an opportunity for people to make money through spamming, it will continue.
The Clearswift SECURE Email Gateway employs a number of different techniques to deliver 99+% detection with minimal administration.
- TRUSTmanager – detects where the message is coming from and, based on the real-time reputation of that connection, can decide whether to accept or reject it. Rejecting the message based on IP connection also means that the actual message hasn’t been sent and no loss of bandwidth has occurred.
- TRUSTmanager is driven by a number of sources of information such as RBL lists, reputation data and the community of other customers’ Gateways feeding back anonymous information about good and bad MTA’s into some special algorithms, allowing Clearswift to score 60m+ addresses every 15 minutes.
- SpamLogic employs a number of spam filtering approaches to sort the different types of spam. Here’s a short description of some of them:
- Greylisting – requires real senders to retry if it’s a sender that’s not been seen before. Botnets don’t typically retry.
- Bounce (NDR) Spam – NDR messages that aren’t generated as result of email your gateway sends are annoying; the BATV filters are developed just for this.
- RBL – allows customers to define additional RBL services if they wish.
- SPF – Sender Policy Framework, more for blocking spoofed messages than spam, but still important.
- Recipient Authentication – By linking into your Active Directory /LDAP directory servers, the Gateway can reject connections to users who don’t exist in your company. Directory Harvesting protection is employed to prevent this being abused.
- Signature based Anti-spam – a hosted signature service ensures that new spam attacks can be detected in near-real time. No need to wait for spam updates before you’re protected.
- Bayesian learning engine – using statistical analysis, this engine can learn about new variants of spam so that if messages do get past all the other filters, the probabilistic engine can detect these.