ARP vs Proxy ARP

[sabih19]

Hi,

Here is case which is making me perplexed;

When I replace a router with another router as a default gateway (same IP but different MAC), the devices become reachable after 1 or 2 packet drops (possibly because of the ARP entry update process).

But, when I replace it with another router with a different IP/subnet and use proxy ARP, Cisco 3750 become unreachable until their ARP tables are flushed. But the non-cisco device becomes reachable after a couple of packet drops (as in Scenario 1). Any explanations!!!

[mcusine]

Hi Sabih,

I think this it's normal due flush timeout in the arp table.

In the case of 6500 the normal behavier it's :

The default ARP table aging time is 4 hours while the CAM holds the entries for only 5 minutes. The switch sends out a frame to all forwarding ports within the respective VLAN when the destination MAC address is aged out from the CAM table. You need a CAM aging timer greater or equal to the ARP timeout in order to prevent unicast flooding. As a workaround, you can issue one of these commands in order to increase the CAM aging timer for the VLAN you are having trouble with to match the ARP aging time:

This is a good explain of proxy arp work.

http://www.cisco.com/en/US/partner/tech/tk648/tk361/technologies_tech_note09186a0080094adb.shtml

Hope this help.

Mauricio

[sabih19]

Thanks Mauricio.

But my confusion remains. When the IP address is changed to a new router (Case1) I don't need to wait for the old def gateway ARP entry to age out on the 3750. Whereas, in Case 2, when Proxy ARP is used to proxy the old def. gateway, the old ARP entry needs to be flushed out before the setup can work as desired.

I we assume that in Case 1, 3750 re-ARPs after the no response to the first couple of packets (seen as ping drops), then the same should work for Case 2. If the 3750 does not re-ARP till the timeout, then it shouldn't have worked in Case 1 too.

[speculor_cisco]

In the first scenario, Cisco 3750 updates the ARP entry for its default gateway 172.25.0.1 when it receives the first ARP request from the new router. In the second scenario, Cisco 3750 adds an entry for 172.25.1.10 but it continues to send the packet to its default gateway 172.25.0.1, without sending an ARP request because it has that MAC address, but with that MAC address the new router discards the frame. When this entry expires, Cisco 3750 will ARP for 172.25.0.1 and the new router will respond doing proxy ARP because it has a route to network 172.25.0.0.

[sabih19]

Hi speculor,

Your point is valid. But, in Scenario 1, pings were initiated from the 3750, and not from the new router. That is why there were a couple of initial ping drops since the 3750 was trying to use the MAC address from the old ARP entry.