Five years ago, I got mad at Barracuda's limitations and went shopping for an alternative. I found lots of expensive options that lacked features which I thought should be the minimum capability of any competent product. So today, I have home-grown tools working that sit in front of my Barracuda. My tools do sender authentication and sender filtering using a rules engine. Barracuda provides content filtering. Barracuda is also a pretty effective tool for message log review
When shopping for a commercial filtering solution, much depends on your situation:
Are you filtering to protect your own organization or are you running an email hosting service that filters email for multiple organizations? If you are protecting yourself, you have more flexibility to decide what is and is not allowed. If your are filtering for others, you have to worry about how much they are willing to pay for the filtering part of your service.
Is your organization willing to put ongoing labor into tuning the email filtering environment, or do you want to give away the problem to a vendor because you have no resources to invest yourself? From listening to news reports, we know that many big organizations are falling victims to ransomware, some of which must be because of email-borne attacks. We are never told which filtering products failed to protect those victim companies! We can bet assume that big organizations are using the expensive cloud solutions, and that they are not a silver bullet against a ransomware disaster.
What is your concept of a "clean" email environment? Perhaps half of the email that we block is nuisance advertising. We block it because every allowed message imposes costs on the user and his organization. It takes him time to decide whether the message is wanted or not, time that he is not spending on his other work obligations. It also costs the organization storage and CPU. If a user has too many nuisance messages, he is likely to miss messages that are actually important to his job. While I can reasonably hope that an outsourcing vendor will block malicious traffic, I don't expect him to understand what constitutes "nuisance" messages to my organization. Maybe you will be able to train the fancy products to your situation, but that means the solution is no longer hands-off for you.
Because SmarterMail attracts budget-conscious organizations, the prior posts on this forum indicate that we use a lot of free or inexpensive tools like Declude, SpamAssassin, and RSpamd.
My custom tools have been around a few principles:
I want to authenticate every message, because all human communication is interpreted in the context of the speaker's identity. If I don't have an accurate identity, then I don't have accurate communication. DMARC and SPF don't get provide enough coverage and don't tell me how to handle exceptions. So I needed to build alternate authentication mechanisms for senders who do not pass SPF and DMARC.
Alternate authentication requires multiple-attribute filtering rules. If an organization has a missing or incorrect SPF policy, a replacement rule would be:
- If HELO name is *.appriver.com", and HELO is verified by forward-confirmed DNS, and MAILFOM name is "*@exmaple.com", then treat the message as equivalent to SPF PASS.
Most commercial products are no better than Barracuda, because they only provide single-attribute rules. In Barracuda, any allow rule requires whitelisting, so the same allow rule is reduced to:
- If the MAILFROM name is "*@example.com", then Whitelist the message whether the message is legitimate or an impersonation.
I never want to whitelist a malicious impersonation, so any ALLOW rule must be tied to an authentication mechanism, an authenticated identifier, and (usually) an identifier that is trusted by proxy because of the authenticated identifier. I cannot do that with single-attribute allow rules, so I cannot make safe ALLOW rules in most commercial products, including Barracuda. Declude and its successor Declude Reboot provide a customizable rules engine, so it became my starting point. But they require an investment of elbow grease.