[threeducks]

Wow, that's smart. I was wondering whether there is a way for the bots to generate "unpredictable" domains such that security researchers could not predict them efficiently (even with source code), but the botnet controller can.

Time-lock puzzles come close, but but it requires that the bots have computing power comparable to the security researchers.

[orlp]

> Wow, that's smart. I was wondering whether there is a way for the bots to generate "unpredictable" domains such that security researchers could not predict them efficiently (even with source code), but the botnet controller can.

There is a fairly simple method which achieves the same advantage for a botnet controller.

1. Use a hash of the current day to derive, for that day, an infinite stream of domain names. This could be something as simple as `to_human_readable_domain(sha256(daily_hash + i))`.

2. A botnet slave attempts to access servers in a diagonal order over (days, domains), starting at the first domain for today and working backwards in days and forwards in domains. An image best describes what I mean by this: https://i.imgur.com/lcEbHwz.png

3. So long as one of those domains is controlled by the botnet operator (which can be verified using a signed response from the server), they can control the botnet.

This means that the botnet operator only needs to purchase one domain every couple of days to keep controlling their botnet, while someone trying to stop them will have to buy thousands and thousands every day.

And when you successfully purchase a domain you can publish the new domain to any connected slaves, so this scheme is only necessary for recruitment into the network, not continued control.

[tea-lover]

Here's the same image on a less horrible file hosting:

https://files.catbox.moe/gilmd1.png

Imgur has been inaccessible for me for months, they're one of those organizations that consider it proper to block whole countries to counter bot abuse.

[stirfish]

Hmm, catbox used to be blocked for me too, but I can access it today. That's interesting.

[itake]

I've definitely heard of cnc using a plural of domains for this reason. the bots have a list of domains they reach out to, searching for one that is valid.

I believe one issue with this strategy is many corporate VPNs block fresh domains. I guess if the software was pinned to use encrypted DNS instead of whatever the OS recommends, then the DNS blocking could be avoided...

[m-s-y]

How would a corporate DNS block new domains, exactly?

[itake]

My employer uses Zscaler. I don't know exactly how they implement this, but my educated guess is the corporate DNS server doesn't resolve domains that were created recently.

In technical terms, the device asks the private corporate DNS server for the IP address of the hostname. The private DNS server checks the requested domain against a threat intelligence feed that tracks domain registration dates (and security risks). If the domain is deemed a threat, either return an IP address which points at a server that shows a warning message (if http traffic) or return an invalid IP (0.0.0.0).

[aaronax]

A firewall. For example, Palo Alto firewalls can easily be configured to block domains newer than ~30 days old.

https://knowledgebase.paloaltonetworks.com/KCSArticleDetail?...

[paranoidrobot]

Have a cache of domains you know about with registration date.

When getting a query for a domain you have not heard about, query whois for it. Store it's registration date in the cache.

[sim7c00]

there are tools pretty good at detecting DGAs these days, but not often implemented.

the best thing to do afaik is use services normal user shave access to, and communicate via those. its hard to tell for anyone who's extracting the data from the third party so the server is hidden. (e.g bot posts images to twitter, and server scrapes the images from twitter, this is also already old news but easier and more likely to sail through that next gen firewall -_-)

i'd say having ur 'own' servers and domains is maybe even a bit dated ( though sadly still very effective!)

[immibis]

It's one of many possible strategies. Any one strategy can be blocked if it's used by enough malicious actors (e.g. Twitter can be forced to block base64 tweets); if they all use different strategies, it becomes harder to justify blocking each individual one.

[sim7c00]

you either need whitelisting, which ppl dont want because they need to send tweets and sync gdrive on their corpo laptops ;')...

so i guess that leaves u with modeling normal user behavior to spot anomalies without the actual packet data being an indicator.

then the bots could piggyback on regular coms still, but it'd definitely raise the bar...

[afandian]

I can see a future where Cloudflare or similar offer a DNS + proxy + Root CA combo to intercept these. Maybe they already do.

[lurkshark]

If I’m remembering correctly, Conficker was the first major use of this technique. They used a relatively small domain pool (250) so the registries were able to lock them up preemptively.

I remember a couple legitimate sites getting slammed by accidental DDOS because the algorithm happened to generate their domain, but having a hard time finding a reference to that.

https://en.m.wikipedia.org/wiki/Conficker

[Tijdreiziger]

Quad9 (the subject of this post) already offers ‘threat blocking’ by default.

https://quad9.net/service/threat-blocking/

[threeducks]

That might work for the current generation of bots, but it will become infeasible when the domain names are generated in such a way that they overlap with spellable and existing domain names.

[0points]

> it will become infeasible when the domain names are generated in such a way that they overlap with spellable and existing domain names.

And why do you believe this will even happen?

[bobmcnamara]

Use a hash chain!

Each time you resolve, the resulting IP can be part of the hash for predicting a future hostname.