ARP Issue in Router, Need Advice

razam

hello,

i have a network setup with one Cisco core switch 4500 series, around 150 access switches that are connected to this single core switch, core switch forwards the traffic to the router 2921.

problem is that the router ARP traffic most of the time reaches 100% utilization and the users will not be able to do browsing, ping will work but there will be no browsing, if i restart the router, it will again start working... it happens around two times in a day..

what is the solution for this ? how can i prevent arp traffic reaching 100% ?

please suggest something.

Mrock4

What version of IOS are you running on your core?

keenon

razam wrote: »

hello,

i have a network setup with one Cisco core switch 4500 series, around 150 access switches that are connected to this single core switch, core switch forwards the traffic to the router 2921.

problem is that the router ARP traffic most of the time reaches 100% utilization and the users will not be able to do browsing, ping will work but there will be no browsing, if i restart the router, it will again start working... it happens around two times in a day..

what is the solution for this ? how can i prevent arp traffic reaching 100% ?

please suggest something.

wow 1 core switch with 150 connected switches. who is routing the 4500 or the 2921? 150 access switches, my guess 48 ports each (just a guess) 7200 mac address entries give or take. I would suggest looking at the specs for your version of ios and mainly the hardware your running. My first thought would be to do a redesign. Add second core switch take routing down to the pair and maybe uplink that router to both but i would really consider replacing the router also ASR100x or 3925/45. Depending on your setup and budget limitations I still would seperate the routing domains core switch routes for access and link from switch to router is out to the wan at least that pulls the arp down to the core and not the router

Mrock4

Whoa! Nice catch. I missed the fact it was 150 access SWITCHES. Thought it was 150 users!

instant000

It looks like all the core currently does is aggregate connections, it is apparently not aggregating networks.

A network redesign can prevent your daily reboots.

The migration plan would look like moving layer 3 as close to the access layer as possible. (Basically, what keenon recommended.)

As far as the core redundancy suggestion, I want to say that it depends.

If you need the availability, having a redundant core might be something you need. (Which keenon also recommended.)

I am not sure if you need that level of availability or not, as you suffer daily reboots as it is, and you still won't have WAN diversity. If most of your critical resources are via the 2921 link to the WAN, making a redundant core might not make much sense. That is, it might be more important to get WAN diversity today, and core diversity tomorrow. of course, if most resources are in the core, then it makes sense to get core diversity today, and WAN diversity tomorrow. So, it just depends.

In many cases, a lot of critical resources may be reachable via WAN, thus increasing the importance of WAN diversity.

Of course, I am just being opinionated at this point. Keenon already gave you a solution.