Friday 24 August 2012

How do ARP flush Entries ?

Its given in RFC 1122


Four mechanisms have been used, sometimes in
                 combination, to flush out-of-date cache entries.

                 (1)  Timeout -- Periodically time out cache entries,
                      even if they are in use.  Note that this timeout
                      should be restarted when the cache entry is
                      "refreshed" (by observing the source fields,
                      regardless of target address, of an ARP broadcast
                      from the system in question).  For proxy ARP
                      situations, the timeout needs to be on the order
                      of a minute.

                 (2)  Unicast Poll -- Actively poll the remote host by
                      periodically sending a point-to-point ARP Request
                      to it, and delete the entry if no ARP Reply is
                      received from N successive polls.  Again, the
                      timeout should be on the order of a minute, and
                      typically N is 2.

                 (3)  Link-Layer Advice -- If the link-layer driver
                      detects a delivery problem, flush the
                      corresponding ARP cache entry.

                 (4)  Higher-layer Advice -- Provide a call from the
                      Internet layer to the link layer to indicate a
                      delivery problem.  The effect of this call would
                      be to invalidate the corresponding cache entry.
                      This call would be analogous to the
                      "ADVISE_DELIVPROB()" call from the transport layer
                      to the Internet layer (see Section 3.4), and in
                      fact the ADVISE_DELIVPROB routine might in turn
                      call the link-layer advice routine to invalidate