from this paper: "First, bandwidth is already a key factor in Tor design. Second, bandwidth is typically a property of a node rather than a link between two nodes, since the bottleneck is likely to be close to the node rather than in the intermediate network."
On Tue, Nov 25, 2008 at 8:46 PM, Daniel Juyung Seo <[EMAIL PROTECTED]>wrote: > Hello, > > Using Tor, I wondered why Tor use bandwidth capacity as a key factor > of Tor router selecting algorithm. > After a brief view of Tor source codes, I realized that the bandwidth > capacity of Tor is more similar to the Tor router's packet processing > capability than the network transfer capacity. > While calculating the bandwidth capacity, the capacity of network link > among Tor routers is not considered. > > Isn't it better to use the capacity of network link among Tor routers? > Robin Snader's paper, A Tune-up for Tor: Improving Security and > Performance in the Tor Network. has some idea using the network link > capacity. > I think it is feasible to measure each link capacity among all Tor > routers by using ping packets or other simple mechanisms. > > When each router knows the linke capacity with all the other routers, > we can build better circuits than before. > Let's assume that there are 1,200 Tor routers in the world. > In this case, each router needs to have 1,199 link capacity > information with the other Tor routers, not the bandwidth capacity of > the other Tor routers. > > The total number of the links that Tor directory server should save is > up to 1,438,800 (1,199 x 1,200). > I guess this mechanism will improve Tor router selection algorithm. > > Waiting for some replies. >
