I was setting up proxies in Kali Linux 2016.2 and was wondering if it is safe(r) to repeat proxies. For example, Proxy 1 > Proxy 2 > Proxy 3 > Proxy 1 > Proxy 2 > Proxy 3 and so on.
-
1What do you mean by "safer"? (1) Less prone to be tracked by your neighbor? (2) Less prone to data corruption? Also, you diagram does not really make sense, if proxies always deliver in a circular fashion you will get nowhere (That's certainly safe for case (1), but not case (2). Since you never get the data back).– grochmalJan 8, 2017 at 22:47
-
I mean would it be harder to trace my IP– DaKouchPotatoeJan 8, 2017 at 22:49
-
if proxies deliver in a circular fashion, it could still come out of proxy 3 after a few repetitions, which is obviously what the OP means– J.A.K.Jan 8, 2017 at 23:01
2 Answers
No, a loop in the proxy chain doesn't increase your anonymity. Rather, it introduces new risks.
First of all, chaining proxies isn't as safe as you'd think. A standard HTTP/SOCKS proxy just forwards the traffic without adding encryption. If that's what you're using, be aware that every proxy in the chain knows
- the address of its preceding proxy,
- the addresses of all subsequent proxies and
- is able to read and modify your traffic (unless you're using end-to-end encryption with the destination server, e.g. by using HTTPS).
This means that even with SSL, all proxies in the chain know which website you're visiting. In contrast, Tor uses an onion routing approach where every node adds a layer of encryption and every hop just knows their immediate preceding and succeeding node. With Tor, only the first hop knows your IP and only the last hop (exit node) can read your original traffic and knows the destination.
Related: Aren't 10 proxies (proxychains) better than Tor with only 3 hops?
Back to your example - consider these cases:
Case 1
you -> p1 -> p2 -> p3 -> https://example.com
Case 2
you -> p1 -> p2 -> p3 -> p1 -> p2 -> p3 -> https://example.com
Imagine that p3
is a rogue node / has to turn over their logs to law enforcement. In case 1, they could only identify p2
as the originator of the traffic and example.com
as the target. In case 2 however, they'd learn about the involvement of p1
, too. From inspecting the traffic, p3
can conclude that it's processing the same traffic twice and would be able to enumerate all involved proxies.
Even if the proxy chain was using Tor-style onion routing, where the traffic wouldn't look the same at the second pass, p3
could still conduct a correlation attack:
you -> p1 -> p2 -> p3 -> pX -> pY -> p3 -> https://example.com
^-- correlation --^
That is, an adversary controlling both ends of a communication channel has good chances of correlating incoming and outgoing traffic by analyzing the timing and volume of the traffic. If you introduce a duplicate node in a chain of proxies, that clearly facilitates a traffic correlation attack and thereby defeats the anonymizing effect of the nodes in between.
(Yay! It's my 100th answer!)
No, this does not add any security, it maybe even removes from it. And it's slower.
It might be worse because there is more communication along that path to you.
identifying you will happen by compromising the proxies or looking at WAN traffic, the first will stay just as hard, the econd will be easier.