Technical Details

You are here: Email Service > Technical Details

SPF + Open Relay database check

SPF (Sender Policy Framework) allows you to check that someone sending an mail is who they say they are. Our mail servers have unique features that combines SPF with grey-listing and the allow mechanism to ensure there is minimal impact on valid mail.
In particular, using greylisting the mail server will to automatically start accepting mail from real mailservers, and using the allow mechanism any valid sender whose message has been blocked is able to add their IP address so that subsequent messages are accepted.

It is important that you have SPF settings correctly configured. If they are not correctly configured you may conclude that SPF is "ineffective" or "stops too much real mail".
This setting may be adjusted on a Per-Account basis... which means you can change your settings to make it looser or tighter.

The mail server's integrated and flexible open relay database checking can be used enforce a servers blacklisting or whitelisting in one or more ORBS databases.

An ORBS database is simply a DNS server that returns a positive response if a server is listed in the database. A variety of services are available online that can maintain blacklist databases. Normally you would maintain your own whitelist database that overrides the blacklist listings.

In addition this can be used to mark messages with a header which can then be taken into account in the Spam "SpamDetect rating" calculation.

Spam Scoring System

This is a message's "spamminess" scoring - based on the sum of the following:

- Locally customisable rule database. Approx 60% accurate on common spam

- Auto training database of recent messages that "look like spam" based on poly and multi symbol statistical word matching. Approx 90% effective

- Auto training database of recent messages that "look like spam" based on message parameters such as URL and content - Approx 40% effective if no local training is done, approx 99.5 effective on trained data.

- Catcher addresses that should never receive genuine mail If mail is received on these addresses it a known a spammer.

- Optional modification of scoring based on ORBS and SPF checks

The auto training databases consist of a base set of rules combined with local training based on messages submitted by the customers as uncaught spam or as a false positive.

Based on this "SpamDetect score" messages can be filtered at a serverwide level or at a per user level allowing individual users to fully customise their filtering & setting up a totally customised "personal antispam policy" based on their chosen level of spam 'tolerance'. One of the more useful techniques is to use the Friends "Challenge / Response" only on messages that look almost certain to be spam.

Custom Filtering rules and Advanced mail rules
A variety of mechanisms exist to setup custom filtering rules, exceptions/overrides to spam filtering and custom mail handling policies at a per user or serverwide level.

Examples of what the per user filtering allows users to is:

- take action (accept / reject / forward / move to special folder) based on message characteristics
- take action (mark/bounce/move to folder/etc) on messages that look like spam
- send challenge response requests to certain messages (look like spam/ from certain domains/etc)

At a domain or serverwide level customised policies can be setup for:

- configuration and customisation of each spam prevention mechanism and default
- mail forwarding, redirection, archiving options
- default per-user rules that are applied if user has no rules defined

- mfilter based advanced mail filtering
If you want to receive or reject any email that has "Model Trains" in the subject line, our system can do it.

Friends Confirmation System

Friends

As mentioned on the other page, the "Friends" System is a spam prevention system that will check incoming mail based upon a list of known good sender email addresses. The Friends system can operate in the following modes:

- Kids safe mode - Only accept mail from known friends
- Keep track of Friends but don't block anything or request confirmation.
- Request confirmation from sender
- SmiteSpam mode - Request confirmation if SmiteSpam gives it a rating bigger than selected.
- Allow all mail through - Friends system disabled

Kids Safe mode
This is the highest level of safety. Mail is only allowed through if the inbound email address is in the list of known friends. If this is not the case the inbound message is bounced with a notification that this person is not accepting mail from unknown senders. Parents can control who their kids are corresponding with.

Keep track of Friends but don't block anything or request confirmation
This mode is most similar to "Request confirmation mode", the difference is that this mode does not hold messages from unknown senders and it does not send a message to them requesting confirmation. So it simply applies the friends list, accepting known friends and leaving unknown ones for another feature/filter to deal with.

Request confirmation mode
In this mode if a message is received from an unknown user an email is sent to this person requesting them to respond to confirm that they are human. If a response is received the original message will be delivered. On a regular basis a status report message is sent to the user providing information of the friends system and any mail pending delivery.

The Friends system may be temporarily disabled if an urgent message is expected from an unknown address using the "Un-Block" button. This will effectively turn either the "Request confirmation mode" or the "Kids safe mode" into "Allow all mail through" for a period of three hours.

SmiteSpam mode
In this mode if a message is recieved from an unknown user and it gets a SmiteSpam rating of more than selected then a request confirmation message is sent as per the 'request confirmation mode' above.