|
I wonder how much botnets reuse IP addresses. Do they infect a system and spam, get blocked, discard the IP and move onto the next (new) one? This means that they have a nearly unlimited supply of IP addresses. Or do they infect a system and spam, get blocked, and then let it go dormant only to awaken it some time later?
I decided to take a look. To do this, I made a list of all the IPs that sent us mail that I could identify as part of a botnet (excluding unknown ones) since July 2011, a time span of 7.5 months. I then took all of the IPs that sent us mail today (or rather, the past 24 hours), and ran them against this list. The IPs that were on the botnet list were eventually blocked as spam. If IPs that sent us mail today appear on this list of botnet IPs, that means that botnets do reuse their IPs.
The question is what is the average time displacement of time-to-block vs time-to-resurrect, and do some bots do it more than others? Here are the results:
Top 10 Botnets that Reuse IP addresses
The numbers in square brackets are relative values compared to the lowest one. This means that if spamsalot reused 20 distinct IP addresses, then cutwail reused 14.5 x 20 = 290 distinct IPs. Or, to put it another way, an IP that sent us mail today was also used by cutwail once in the past 7.5 months. By contrast, 14.5 IPs that sent us mail today were used cutwail in the past 7.5 months.
In other words, cutwail reuses its IPs much more than any other botnet.
Digging deeper, and explaining it better because the above paragraph makes no sense, I took a look at where the biggest amount of reuse occurred. Of all the IPs that sent us mail today, that at one time sent mail for cutwail, 28% of them were on Oct 27. The next largest value, Dec 17, only comprises 3%. Thus, for cutwail, there is a lag of around 3.5 months before it tries to reuse the same IPs. Asprox shows a similar reuse pattern for Oct 27.
Thus, while botnets do reuse IPs to send spam after they have been blocked, they don’t do it a lot. Of the ones that do resend, they don’t use very many of them again.
Clearly, spammers know when they have been blocked and keep bringing up new troops.
Sponsored byVerisign
Sponsored byDNIB.com
Sponsored byIPv4.Global
Sponsored byWhoisXML API
Sponsored byCSC
Sponsored byRadix
Sponsored byVerisign
Did you look and see whether listings on botnet BLs, notably the CBL, affect botnet reuse? I’ve always assumed that botnets look and see where they’re blocked, so they can sell “fresh” bots for more money, but never had enough data to check.
My data mapping of IPs does come from the CBL. I haven't managed to do a fuller analysis yet because the CBL data is extremely large. But that would be interesting. We lose data of IPs once they end up on a blocklist. One day we'll get it and I'll be able to do an analysis but my initial guess is that the patterns would be similar to the above.