2

I have just finished reading the posting on arma's blog about entry guards and I'm trying to reconcile what I have read there with what I (think) I understand from other postings about Tor.

First, I assume (please tell me if I correct or not) that a new set of guards is selected each time a user initiates a Tor session, so the remarks about retaining guards for periods of 6 to 12 months would suggest that some Tor sessions are of extremely long duration. Is that correct, and are there any statistics about the distribution of duration of Tor sessions?

Second, there are several warnings, in other Tor guides, not to mix different identities in the same Tor session. I understand that to mean that, if I use a pseudonymous identity of Frank in one session and access websites that recognize me as Frank, then I should not also use the pseudonymous identity of George (to access the same or different websites) in the same session. Again, is that correct?

Third, how does one reconcile the apparent desirability of maintaining guards (and, therefore, presumably a session) with the "don't mix identities" warnings.

I would very much appreciate some clarification both of whether my understanding is correct and of the apparently competing risks/benefits. Alternatively, if there are particular documents that I should read to clarify a misunderstanding or to answer my own question, then I would appreciate a pointer to those docs.

4

Third, how does one reconcile the apparent desirability of maintaining guards (and, therefore, presumably a session) with the "don't mix identities" warnings.

I believe you are mixing two concepts here. Guard relays protect your connection to the Tor network. The "identities" (Frank and George) are presumably id's on web pages, and as such only visible in traffic leaving the Tor network from the exit relay.

Tor maintains your guard relays automatically. Every few weeks new guard relays will be selected. No need to think about it.

Don't mix identities in a session probably refers to the fact that during one session you will use one exit relay. An opponent watching the traffic from the exit relay could figure out that Frank and George are connected. To solve this Tor provides a means of switching to another exit relay, the "Use a New Identity" button in Vidalia.

  • I get the distinction between persistent entry guards, which are only observable locally, and identities seen by websites (and for unencrypted traffic, by exit relays). However, for users who access Tor in multiple environments, don't persistent entry-guard choices fingerprint them? How many entry guards are there? Given that, one could estimate the uniqueness of a given choice. Maybe this is a separate question ;) – mirimir Oct 26 '13 at 4:45
3

First, I assume (please tell me if I correct or not) that a new set of guards is selected each time a user initiates a Tor session

That's incorrect. Tor stores the details of the guard nodes it used during its last session. Unless the appropriate time period has elapsed, it will reuse the cached ones when it next starts up.

Tails, due to its non-persistent nature does note take advantage of this, as it has no way to save the guard details between reboots.

  • Indeed, that's how entry guards can be persistent. What OP describes is the current situation for Tails. – mirimir Oct 25 '13 at 1:14
  • that's true - although OP didn't mention Tails, I'll add in a note about it. – Megan Walker Oct 25 '13 at 11:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.