9

Does using the pluggable transport Tor Browser Bundle with an Obfs3 bridge affect browsing speed in a noticeable way?

I am interested in using an Obfusproxy to hide my Tor usage form my ISP but I want to evaluate the effect on browsing speed.

I understand that there are restrictions imposed by the bridge themselves (thanks alaf for that comment) but assuming that the obs bridge is equivalent to any guard node that I may connect to, how is my local resource different for obfs bridges versus regular Tor usage

2 Answers 2

3

When ignoring the factors pointed out in alaf's comment, we are left with the actual protocol differences. The obfs3 protocol has both fairly low computational and communication overhead. The computational overhead stems from the use of AES-CTR-128 which wraps Tor's already-encrypted network traffic. The communication overhead is caused by the variable-sized handshake which prepends the data transfer. The handshake consists of 192 bytes for the public key, on average 4097 bytes of padding and a 32-byte HMAC. That sums up to an average overhead of 4321 bytes.

In practice, there is some additional computational overhead caused by the data stream going through obfsproxy.

In conclusion, the difference is probably not noticeable unless you are on a platform with highly constrained resources. Note that while obfs3 is fairly efficient, other pluggable transports provide better obfuscation at the cost of lower throughput.

3

There is no absolute answer to this question.

Your browsing experience, meaning mostly the lag to open or browse a web service, is depended on the resources of the obfuscate bridge you connect to. We can assume with fair certainty that some obfs bridges have more resources than other.

What's more, your experience will vary due to variance of usage of the obfs bridge in use, meaning if many users are exhausting its resources.

Finally your browsing experience will be degraded if one or more of the obfproxies you've chosen go offline. In that case you are not as flexible as when you connect to ordinary guard nodes. Instead, you have to fetch more obfs bridges' IPs and use them as entry nodes.

An insight on how Tor Project is distributing obfs bridges to users could be helpful but again obfs bridges and normal bridges is part of an arms race. Tor Projects needs to distribute bridge IPs to good people while keeping them secret from bad people. That makes public info on obfs bridges less accessible.

Generally, my understanding is that bridges and obfs bridges are fewer than needed and an average bridge has less bandwidth than an average guard node.

Of course a bridge with lower advertised bandwidth should be used by/distributed to fewer users, but that's not entirely controllable since it involves many parameters.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .