Email is inherently a malicious traffic stream

It’s something many people don’t think about, but the majority of the traffic coming into the SMTP port is malicious. Spam is passively malicious, in that it just uses resources and bothers people. But there is a lot of actively malicious traffic coming into the SMTP port. Email is used as a vector to spread viruses and other malware. Email is also used for phishing and scamming. Many of the major hacks we’ve heard about over the last few years, including those in the email space, started with a single user getting infected through email.
We talk a lot about delivery here with clients and primarily focus on making sure their mail looks as unlike malicious mail as possible. We focus on spam filters, but every piece of mail goes through filters that also look for viruses, phishes, malware and other malicious traffic.
Mail servers are under attack constantly. The only reason our inboxes are useful is through the hard work of many people to filter out the bad and keep users from seeing the bulk of the mess attacking them.

Related Posts

Does email have a guarantee of delivery?

A client asked me earlier this week what SLAs ISPs provided for email delivery. The short answer is that there isn’t a SLA and that the only guarantee is that the email will get there when it gets there.
But as I was mentioning this to Steve, he pointed out that there was a recent change in the RFCs for email. In both RFC 821/2 and RFC 2821/2 (the original email related RFCs and the update in the early 2000’s) the RFCs stated that once a receiving MTA accepted an email that that MTA was required to either delivery the mail or generate an asynchronous bounce. While this isn’t a standard SLA, it does mean that a 2xy response after DATA meant the email would either be delivered to the user or be sent back to the sender. Despite the RFC requirements some receivers would still drop mail on the floor for various reasons, sometimes intentionally and sometimes not.
RFC 5321/2, the current SMTP standard, still says that once a server accepts the mail it must not lose that mail ‘for frivolous reasons.’ The RFC goes on to admit, though, that in recent years, SMTP servers are under a range of attacks and dropping mail on the floor is not frivolous in those cases.
 

Read More

The best time to send email

This subject comes up over and over again. Many senders are convinced clock_at_sign that there is a best time to send email. Countless research hours have been dedicated to finding that best time to send email. Numerous blog posts discuss what the best time to send email is.
From my perspective, there are better places for senders to spend time than figuring out what the exact right time is.But, senders still ask when the best time to send mail is.
There are a lot of reasons I can come up with as to why there’s no best time to send email. But the really big one is that when you send a mail has no impact on when it gets delivered.  There are multiple steps between hitting the send button and the mail being delivered to the inbox totally outside the control of the individual sender.
Email is designed as “store-and-forward.” This means there are potential delays at multiple steps inside the process.
Sending queues are called queues for a reason. Emails are sent out individually, particularly when an ESP uses VERP as part of its sending. There is actually a time overhead for making a connection to a recipient server and sending the email.
Receivers have queues, too. They can only accept so many incoming connections at a time. They have limited resources to accept all the mail their users want.
Receivers may delay mail between accepting it at the MX and delivering it to the inbox. This isn’t ideal and it’s not usual, but it can happen.
Recipients using IMAP accounts may not check mail regularly. They may only collect mail a few times a day.
These are only a few of the reasons that send time doesn’t necessarily equate with delivery time. Of course, 99% of the time email is mostly instantaneous. The internet is robust enough that a message sent is delivered seconds later. I see it happen all the time, when colleagues and I send email during calls. But, when mail fails, it sometimes fails spectacularly. Back in the dark ages (of the early 90s) I had an email that took almost a year to get to the recipients. Best I can tell, it got stuck somewhere in the depths of a machine in the middle of the university mail system. Eventually that system fell over and someone noticed and rebooted it (maybe it was walled up somewhere?).  The reboot shook my message out of where ever it was stuck.
 

Read More

SMTP Level Rejections

While discussing a draft of a Deliverability BCP document the issue came up of what rejections at different phases of the email delivery transaction can mean. That’s quite a big subject, but here’s a quick cheat sheet.
At initial connection
Dropped or failed connection:

Read More