Proc Pegged at 100%

[Live2 Bicycle]

Friday our server guy decided to start coping 200 gig worth of data from our primary centera server to the offsite centera located at another of our facilities. With in minutes of him starting this transfer I got many calls from the remote site stating the network had come to a crawl. I verifed by doing a sho pro which showed the rtr pegged. I put ip cach flo on all the intfaces then did a show ip cach fl.

Fa0/1 192.168.25.42 Fa0/0 192.168.20.42 06 C83E 0C92 1162K

Fa0/0 192.168.20.42 Fa0/1 192.168.25.42 06 0C92 C83E 252K

Fa0/0 192.168.20.42 Fa0/1 192.168.25.42 06 0C92 C551 251K

Fa0/1 192.168.25.42 Fa0/0 192.168.20.42 06 C551 0C92 1922K

Wha lah I was able to determine the two boxs killing the rtr. The server guy tells me this is now going to be an ever day occurance from now on. The two sites are connected via a 100 microwave. The remote site has a 2621 rtr with a Fast Ethernet port to the microwave and the local LAN. At the main site the microwave is connected to a 3548XL switch.

What is my best solution to help the server guy back up his data to the offsite location without killing the network for the user base at the remote facility? Should I replace the rtr with a layer 3 switch? Should I use rate limit and QOS on the rtr? Thanks in advance for any feedback.

[tekha]

I suppose you have allready enable ip cef.

According to this pdf "Router Performance"(http://www.cisco.com/warp/public/765/tools/quickreference/) , a 2621 can only handle around 13 Mbps and a 2621XM around 15 Mbps. So if the bandwidth of the microwave connection is more then that, it makes sence to change the 2621 to something faster, maybe a layer 3 switch.

[jennie927]

---

@tekha wrote:

I suppose you have allready enable ip cef.

According to this pdf "Router Performance"(http://www.cisco.com/warp/public/765/tools/rushroyale/ ) , a 2621 can only handle around 13 Mbps and a 2621XM around 15 Mbps. So if the bandwidth of the microwave connection is more then that, it makes sence to change the 2621 to something faster, maybe a layer 3 switch.

---

I agree with this guy. You probably need a faster device.

[Joseph W. Doherty]

Ah, that's an interesting reference - the "rushroyale" part.

You probably intended to reference the attachment I provided.

If so, understand many of those listed Cisco routers can often perform better as performance stats are for 64 byte packets, and often most traffic averages packet sizes represented in/by IMIX.

However, 100% CPU is often caused by hitting the physical packet forwarding performance limits of the router.

Also, in this original case, there's still likely to be a perceived "slow network" problems whenever a link is continuously driven at 100% utilization too.   Again, for the OP case, QoS might be very good for mitigation.

[Wendllandsonlamont707]

Have you fixed the issue. As we are also facing the same issue in the router of our office. Where we are providing stickman warriors download services to our clients. But from the last 2 weeks due to this issue of the router we haven't provided any services to our clients and due to which we are lossing our clients. How you fixed this issue? 

[Joseph W. Doherty]

As @tekha described (17 years ago), if you exceed the forwarding capacity of a device, in particular the OP's 2621 router, you'll need a "faster" device.  All these years later, that's still true, although current routers and switches have much more capacity than they did back then.

Not touched upon by @tekha, routers or switches with sufficient capacity can often saturate a network link, and without QoS (asked by @Live2 Bicycle as a possible mitigation option), that too can be perceived by users as "the network had come to a crawl."

@Wendllandsonlamont707 I suggest you create a new posting further describing your network environment and your network issue.