I have been running load time tests for duckduckgo.com and ff.duckduckgo.comthrough the Tor network.
duckduckgo.com should be on an enclave nodehttps://www.dan.me.uk/torcheck?ip=72.94.249.36, ff.duckduckgo.com should not be. But they appear to be loading at comparable times and LoadUI is showing no difference between the two URLs- even more complex queries such as "what is my IP"
Can anyone tell if duckduckgo.com misconfigured? Is it the CDN that DuckDuckGo uses that equalizes the load time? Is the Tor network not utilizing the enclave node as heavily as the documentation makes it sound? Or have the exit proxies really gained that much capacity that they are no longer the bottleneck for such small small requests?
-Zach Lym
On Thu, 19 May 2011 01:14:30 -0700 Zach Lym indolering@gmail.com wrote:
I have been running load time tests for duckduckgo.com and ff.duckduckgo.comthrough the Tor network.
duckduckgo.com should be on an enclave nodehttps://www.dan.me.uk/torcheck?ip=72.94.249.36, ff.duckduckgo.com should not be. But they appear to be loading at comparable times and LoadUI is showing no difference between the two URLs- even more complex queries such as "what is my IP"
Can anyone tell if duckduckgo.com misconfigured? Is it the CDN that DuckDuckGo uses that equalizes the load time? Is the Tor network not utilizing the enclave node as heavily as the documentation makes it sound? Or have the exit proxies really gained that much capacity that they are no longer the bottleneck for such small small requests?
Tor finds an exit enclave for a server by looking for a relay whose IP address matches the destination server's IP address. Thus, Tor cannot begin to use the exit enclave until it has performed a DNS lookup for the server's hostname. Tor will also send the first few streams to a server with an exit enclave over an existing circuit; an exit enclave is only used once Tor has built a circuit to the enclave relay.
Robert Ransom
I have been running the LoadUI tests for over 2 hours, I don't think they produce a single TCP stream.
I geting almost no difference between out-proxy, enclave, and the hidden service versions of the site. The LoadUI 1.5 project file is herehttp://www.indolering.com/wp-content/uploads/2011/05/TorBeta.xml_.zip. if you care to check.
Does the majority of the network use an exit enclave if one is available?
Thanks, -Zach Lym
On Thu, May 19, 2011 at 9:13 AM, Robert Ransom rransom.8774@gmail.comwrote:
On Thu, 19 May 2011 01:14:30 -0700 Zach Lym indolering@gmail.com wrote:
I have been running load time tests for duckduckgo.com and ff.duckduckgo.comthrough the Tor network.
duckduckgo.com should be on an enclave nodehttps://www.dan.me.uk/torcheck?ip=72.94.249.36, ff.duckduckgo.com should not be. But they appear to be loading at comparable times and LoadUI is showing no difference between the two
URLs-
even more complex queries such as "what is my IP"
Can anyone tell if duckduckgo.com misconfigured? Is it the CDN that DuckDuckGo uses that equalizes the load time? Is the Tor network not utilizing the enclave node as heavily as the documentation makes it sound? Or have the exit proxies really gained that much capacity that they are
no
longer the bottleneck for such small small requests?
Tor finds an exit enclave for a server by looking for a relay whose IP address matches the destination server's IP address. Thus, Tor cannot begin to use the exit enclave until it has performed a DNS lookup for the server's hostname. Tor will also send the first few streams to a server with an exit enclave over an existing circuit; an exit enclave is only used once Tor has built a circuit to the enclave relay.
Robert Ransom
tor-relays mailing list tor-relays@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays
On May 19, 2011, at 9:58 PM, Zach Lym wrote:
I have been running the LoadUI tests for over 2 hours, I don't think they produce a single TCP stream.
I geting almost no difference between out-proxy, enclave, and the hidden service versions of the site. The LoadUI 1.5 project file is here. if you care to check.
Does the majority of the network use an exit enclave if one is available?
Thanks, -Zach Lym
There should be no (big) difference in speed, because tor uses the exit enclave as the fourth hop. So you have a "normal" three-hop circuit which gets extended to the enclave.
Sebastian
So an enclave node doesn't bypass exit nodes, or are exit nodes no longer the bottleneck they once were?
My hypothesis was that if a major service providers provided an exit enclave, traffic to them would bypass the exit node bottleneck. Given the distribution of internet traffic (the top 10 sites account for 20% of traffic to the top 1,000 sites http://tinyurl.com/3agt69j) tor could see some serious performance gains.
Thanks, -Zach Lym
On Thu, May 19, 2011 at 1:22 PM, Sebastian Hahn mail@sebastianhahn.net wrote:
On May 19, 2011, at 9:58 PM, Zach Lym wrote:
I have been running the LoadUI tests for over 2 hours, I don't think they produce a single TCP stream.
I geting almost no difference between out-proxy, enclave, and the hidden service versions of the site. The LoadUI 1.5 project file is here. if you care to check.
Does the majority of the network use an exit enclave if one is available?
Thanks, -Zach Lym
There should be no (big) difference in speed, because tor uses the exit enclave as the fourth hop. So you have a "normal" three-hop circuit which gets extended to the enclave.
Sebastian _______________________________________________ tor-relays mailing list tor-relays@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-relays
tor-relays@lists.torproject.org