Advice re: Unicast flooding and ARP timeout

[cbeswick]

Hi,

All our distribution switch blocks utilise dual 6509 chassis' / sup 720 modules. We load balance gateways and traffic by using HSRP and RSTP, nothing new here.

We seem to be hitting a common issue of unicast flooding which I now understand is a direct result of the cam tables timeing out every 5 minutes.

I plan on reducing the ARP timeout for all our SVIs to 480 seconds to resolve this issue.

My question is this: When I change the ARP Timeout value, are all the existing ARP entries in the given SVI flushed, or does the timer just reset to 480 seconds and re-poll entries when the aging timer expires ?

I am trying to understand the impact this change may have on live services, especially on the Server distribution layer.

Thanks in advance.

Chris.

[simontibbitts · Accepted solution]

Hi Chris.

I just tried changing the ARP on an interface of a router and it made no impact. It looks like when you change the ARP timeout all new ARP entries will be given that new timeout. The ARP entries currently in the ARP cache will continue with their current ARP timeout values until they timeout and then they will receive the new timeout value.

So no, changing the ARP timeout will not cause an issue.

Just in case you have not seen it this link is relevant to your problem. It is a little old now but still very valid.

http://www.cisco.com/en/US/products/hw/switches/ps700/products_tech_note09186a00801d0808.shtml

Simon

[cbeswick]

Hi Simon,

Many thanks for your response / cofnirmation.