Most of us have heard of the HTL=1 attack to determine whether or not a
file is stored on a node.  This was addressed, albeit lamely, by adding
a probabilistic factor so that there is still a 22% chance that the node
will forward an HTL=1 request.

Of course, in reality if the node responds with the data quickly you
know it was caching it, and if it takes one hop time, you know it
wasn't.  This is true for requests made at any HTL.  A fast reply means
the data was cached at the node.

I propose that if the node finds the data for a key in its cache, or if
it's the end node in the chain, it ALWAYS waits one hop time (chosen
randomly within the statistically correct range) before sending the
DataReply or InsertReply/DataNotFound.

-tc
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 240 bytes
Desc: not available
URL: 
<https://emu.freenetproject.org/pipermail/devl/attachments/20020208/e513bb47/attachment.pgp>

Reply via email to