I'm having trouble understanding parts of the original Kademlia paper
(http://pdos.csail.mit.edu/~petar/papers/maymounkov-kademlia-lncs.pdf).
Can anyone provide some clarity around the following sections:
Cache expiration time
---------------------
In section 2.3, what does the paper mean by "exponentially inversely
proportional"?
Routing table
-------------
The last paragraph of section 2.4 seems to state that all valid contacts
should be kept in the routing table. If the k-bucket for the prefix of
those contacts is full, the k-bucket should be split in to 2 and moved
down 1 level in the tree. So, for example, if the k-bucket for prefix
001 is full, then it should be split in to 2 k-buckets: one k-bucket for
prefix 0010 and another for 0011.
However, previous sections seem to state that if the k-bucket already
has k elements, that any further additions in to the bucket require
removing the stalest node (pinging it first to see if its alive).
So which is correct? Should the bucket be split in to 2 or should the
bucket reject the addition if no more room is available?
_______________________________________________
p2p-hackers mailing list
[email protected]
http://lists.zooko.com/mailman/listinfo/p2p-hackers
