I had a VZB T3 terminating on a 2851 with NM-1T3/E3, sufficient mmory and an AIM-VPN card.
T3 came up fine after initial installation, however, after a few minutes, performance degraded significantly. There were no errors on serial or controller. VZB also checked out line clean on their side.
After rebooting router, the same behavior was seen. Right after reboot, it works fine but after a few minutes, it slows down significantly. Again, no errors on serial or controller.
The same card on a Cisco 3725 does not have any problems.
Cisco 2800 Series aren't powerful enough to handle T3 full-rate. They are powerful enough to handle sub-rated T3 WAN connections.
As you noticed, the comparable model for a 3725 would be a 3800 Series router, thus you need to look at that router as an ideal replacement for your 3725.
A 2800 Series router is comparable to the 2600 Series.
I understand that 3800 is comparable repalcement for 3700. However, I have seen 2851 used for T3. If I remember correctly, 2851 has similar or better horsepower as 3725.
I am waiting for the router to be shipped back to me from NY. I plan on re-deploying it locally so I can T/S. Will post them once I have router back.
Please refer to the Module Data Sheet (Table 1) which reinforces what I said in my first post:
Horsepower numbers are often taken from LAN-to-LAN connections.
Please rate helpful posts
CEF performance for a 2851 is noted as 220 Kpps, for a 3725 as 100 to 120 Kpps (ditto for 3660). It would seem the 2851 should offer similar or better performance vs. the 3725. So, it's curious the 3725 works fine with the T3 card but the 2851 doesn't.
Are you using the NM-1HSSI or the NM-1T3/E3? What IOS are you running on both the 2851 and 3725?
BTW: I don't discount Edison's posts, but it still seems odd that a 2851 with 2x the CEF pps rating of a 3725 would struggle with a T3 when the 3725 doesn't. Of course, it's certainly possible there's more to the equation than rated PPS performance. It might also be possible that something else is an issue, for instance if your traffic is causing processing switching on the 2851 but not the 3725.
He said NM-1T3/E3 on his original post.
The URL I posted indicates how the module is rated according to the hardware used.
The pps/bps figure are often measure in LAN-to-LAN connectivity. For WAN support, I always recommend checking the hardware specs on the WAN module for their bandwidth support on the intended hardware.
On this case, if the OP was to call TAC, he will be told that's normal as the specs position this hardware for subrate T3, not full rate.
"He said NM-1T3/E3 on his original post."
He certainly did! How did I miss that? Oops!
"The pps/bps figure are often measure in LAN-to-LAN connectivity."
Agreed, and although LAN vs. WAN can be apples vs. oranges, I'm still surprised there would be such a difference between the 2851 and 3725 platforms, although it's certainly possible.
Within your later post's reference, you find:
"Q. Does the Cisco 2800 Series support a T3/E3 interface?
A. The Cisco NM-1T3E3 is supported on the Cisco 2811, 2821 and 2851 platforms. This module provides customers with a connectivity option for a DS3 interface, but only at sub-line rates. As shown Table 1, the Cisco 2800 Series routers are designed to be T1/E1 routers, not DS3 routers. Although some customers may achieve line rate or near line rate on a DS3 with no services, the addition of common services such as access control lists (ACLs), quality of service (QoS), and IP Security (IPSec) will each cause performance degradation. The recommended platform for T3/E3 connectivity is the Cisco 3800 Series."
We don't know the config, but once again, I would expect a 2851 being able to "achieve line rate" on a T3. Besides, I've worked with a 2811 with a full T3. No, it doesn't achieve line rate, but with extensive ACLs and QoS it maxs out at about 50% (duplex) of the T3. Since the 2851 is about 2x the performance, I would expect it almost fully push the T3.
"On this case, if the OP was to call TAC, he will be told that's normal as the specs position this hardware for subrate T3, not full rate. "
No doubt, so I inform customers Cisco ISRs can often deliver performance beyond the Cisco recommendation, this is one of the risks when doing so. Not something to be discounted lightly.
I don't remember where I got the attached PDF, but I use it as a reference for router performance. It shows the 2851 should be able to move 112Mbps (at 64byte packets). Obviously, there are many factors that influence router performace beyond speeds-n-feeds.
I agree. I also do not discount Edison's notes but continue to question why 2851 would suffer.
2851 is running 12.3(11) Adv IP Serv.
3725 is running 12.4(17) IP Base.
I know for certain that router is not configured for process switching. Doesn't router use CEF by default?
I guess I have to provide another link?
Again, the numbers posted for performance in the 2800 or any other typical router are for LAN-to-LAN.
Always check the documentation for module support and limitation before deploying...
I understand your point. At the same time, this is not first 2851 I have deployed with a T3 which is why I raised my original question.
I will re-deploy it with the same connectivity once rotuer gets back here and post findings.
Thanks for yor help/input.
Here is what I found.
interface Serial1/0.1 point-to-point
ip address x.x.x.x
no ip redirects
no ip proxy-arp
snmp trap link-status
no cdp enable
no arp frame-relay
frame-relay interface-dlci 500 IETF
The above was configured on the interface.
Notice "Class VOICE". I was told that "map-class frame-relay VOICE" however was not configured. This caused BW to default to 56K.
The router currently is in service and is performing as expected.
Thanks for your help.
Yes, CEF should be normally be the default, but what might cause a packet to be processed switched, I believe, can vary based on IOS. Later IOSs seem to continue to decrease the process switch conditions.
I also see your later post. If you've used other 2851s with T3s, same IOS?
If the 2851 is under maintenance, you might try the latest version of 12.3 or 12.4.