TCP session disconnection caused by Code Red?

RFC 1122 mandates that you query for a particular ARP
destination no more frequently than once per second.

RFC 1122 also notes a number of reasons why people may want to make
the positive ARP cache timeout long -- if one suppresses ARP queries
for that time, you'll have the situation where if a popular host goes
down for a period of time, it is effectively off the network for a long
period while waiting for ARP negative caches to timeout. Probably
a bad idea. Rate limiting, as RFC 1122 suggests, would seem to be much