Cisco Router 1941 Serial Overrun Issue

[Manuel Cruz]

Hi Everyone,

I've been working with our companies' 2 Cisco Router's for about 1 1/2 years now and i've come across one issue i havent been able to remedy or find a viable solution for short of purchasing 1 model or 2 up. Our 2 Cisco routers handle our voice traffic between 2 sites and those that work at home. The 1941 in my office has no issues at all. The 1941 in my datacenter where the phone server is located is the one that is displaying issues. This one also provide our at home users with voip access to use our phones at their house. Each of these routers contain the standard 2 LAN ports GE 0/0 and 0/1 and a CSU/DSU card that connects the 2 routers over a P2P connection which is a 1.5mbps connection. The only interface that displays the overruns is the serial interface on the datacenter side. I looked at all counters on my office side and i see nothing that would show any sign that the traffic would be coming from my side. I then go back to the other interface on the problemmatic router and see that the GE 0/0 interface that faces my phone server has increasing numbers in pause outputs. At first i thought this was the result of something else but now it seems clearer that the issues in its entirety exists on the router itself. The traffic is moving too fast between the GE 0/0 and the Serial 0/0/0 to the point that the interface is telling the other to slow down which in turn adds to the pause output counter. The counter being added to the overrun is just another bi-product of the bulk in traffic being pushed towards the P2P before it makes it over it. So I'm wondering whether or not changing the queueing strategy might help or modifing the hold queue or going as far as manipulating the buffers which i wouldnt do regardless. Most of my research has shown me that its the router that needs to be upgraded when its the interface's on board memory that can only handle so much reagrdless of whether i change out the interface to an identical one that handles the same amount of traffic over the P2P.

I appreciate any and all help/advice my fellow IT community can provide.

PS. I've attached screenshot of datacenter router 2 interfaces in question.

[Manuel Cruz]

I'd like to add that i found out this had happen to people when replaced a switch that was connected to the router with a higher performing one. This issue started after replacing the Netgear switch i previously had with a Juniper EX model that out performed it on many levels and scenarios. My search showed that the Juniper was moving and queueing traffic at a faster rate then the Netgear which was causing the GE 0/0 interface which is on a FiFo queueing stratedgy to move traffic at a faster rate to the Serial 0/0/0 interface which can only move the traffic at a given rate cap of 1.5mbps over the point to point. I was thinking of changing the juniper interface MTU to 1500 which matches both of the interface on the Cisco Routers. What do you guys think?