[e2e] Transparent subnets with Proxy Arp

[Bob Braden]

  *> 
  *> What was so bad about using proxy ARP to implement transparent
  *> subnets a la RFC 1027? (aside from being a violation of the e2e 
  *> principle)

Who says it is "so bad"?

Section 2.3.2 of RFC 1122 seems to pretty clearly lay out the dangers
-- proxy ARP can result in much more frequent ARP cache invalidation,
which has implications for ARP overhead and perhaps packet loss.
What is it about Section 2.3.2 that you don't understand?

  *> 
  *> The Host Requirements RFC includes a comment about it 
  *> interfering with transport layer dynamics.  And some other

Please give a specific reference; I don't see that comment.

Bob Braden

  *> references mentioned problems when an ARP table maps
  *> several hosts to a single MAC.  In both cases the actual
  *> problems weren't described.
  *> 
  *> Any help on this would be much appreciated.  Thanks!
  *> 
  *> - Eric Goldman
  *> 
  *>