dtnrg-4----Page:17
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
|
There is concern that detecting replays at arbitrary nodes would require each node to maintain an excessive amount of state Some replays, such as retransmitted bundles and replicative routing information are legitimate; additional bundle headers would be required to distinguish legitimate from illegitimate replays DoS attacks as executed by replay attacks are expensive and difficult to mount versus other types of DoS attacks, and they would be costly to protect against, so it does not make sense to incorporate mechanisms for defending against them By explicitly not defining mechanisms to detect replays, we may be leaving some networks that have bandwidth constraints vulnerable to unintended routing loops; to protect against unintentional routing loops, we probably want a mechanism for detecting and discarding bundles that circulate excessively in the network Why replay detection is not provided at every hop |