I am not very well versed in frame relay and in trying to troubleshoot some speed issues at our remote locations, I have the following questions.
I am trying to understand the specific of the frame relay circuits we have installed for our remote locations. For instance, we have 128k circuits from our telco provider at a handfull of remote locations. They other end of the circuits is the DLCI here at our headquarters. I have several sites complaining about speed issues and in looking into the problem, I saw the following:
Concord_3662#show frame-relay pvc 125
PVC Statistics for interface Serial1/1 (Frame Relay DTE)
DLCI = 125, DLCI USAGE = LOCAL, PVC STATUS = ACTIVE, INTERFACE = Serial1/1.125
input pkts 12465221 output pkts 10663317 in bytes 1503144387
out bytes 2807619455 dropped pkts 5219 in pkts dropped 0
out pkts dropped 5219 out bytes dropped 1918626
late-dropped out pkts 5219 late-dropped out bytes 1918626
in FECN pkts 0 in BECN pkts 26460 out FECN pkts 0
out BECN pkts 0 in DE pkts 0 out DE pkts 0
out bcast pkts 867686 out bcast bytes 72616279
5 minute input rate 3000 bits/sec, 5 packets/sec
5 minute output rate 56000 bits/sec, 6 packets/sec
pvc create time 6w1d, last time pvc status changed 3w1d
cir 56000 bc 7000 be 0 byte limit 875 interval 125
mincir 28000 byte increment 875 Adaptive Shaping none
pkts 2296509 bytes 672679388 pkts delayed 614623 bytes delayed 454832630
traffic shaping drops 0
Queueing strategy: fifo
Output queue 4/40, 5262 drop, 614623 dequeued
I am most confused by the CIR, MINCIR, BC and BE counters. If we have a 128k circuit and the CIR is 56k with MINCIR of 28k, BC (committed burst?) of 7k, and BE (excess burst?) of 0, what are my guaranteed and realistic speed going to be on this circuit?
Any assistance would be greatly appreciated.
Thanks in advance.
Frame-relay traffic shaping is enabled on the main interface, and it applies to all DLCIs under that interface.
If a certain DLCI has no map class attached to it, and traffic shaping is enabled on the main interface, the DLCI is assigned a default map-class with CIR = 56000.
This should explaint it...(watch out for the wrap)
OK, that explains some of this then. We attempted to turn on that feature to prioritize Citrix traffic over other applications for one location. If it worked out well, we were going to look into doing for others as well.
Can it be enabled on just a DLCI and not the main interface? Secondly, I was told that all routers would have to be running at IOS level 12.2 to use the traffic shaping features. All of the other routers are at 12.1 so we are not sure if we do need to upgrade or not. If you have experience with this, can you confirm whether or not this is true.
I notice from your output that there are HIGH BECN on the circuit, this suggests to me that there is congestion somewhere on the Frame Network, my question to you would be have you spoken with your FR provider to investigate this?
Also, heres a quick explanation on BECN + FECN
Flow control is an important function in all types of network, and frame relay is no exception. The purpose of flow control is to prevent the originating party from sending more information than the receiving end can accept. Sometimes a specific part of the network is overloaded because it cannot keep up with the overall pace.
If the network assumes the role of the receiver of information in these situations, flow control can be used to avoid serious degradation of network performance. This is where the BECN and FECN bits come into play. When a frame-relay node has been overloaded, all BECN and FECN bits in all frames on all connections through the node are flagged.
The purpose of the BECN and FECN bits is to inform the end-users of this overload. The party that is responsible for the transfer of information (either the sender or the recipient) must then reduce the transfer rate to allow the network to recover.
Unfortunately, only a few end-user protocols can interpret the BECN and FECN bits.
The TCP/IP protocol is designed for equally effective use on leased lines and in X.25, ATM or frame-relay networks. In other words, it should perform the same functions, regardless of which lower-layer protocol is used. That is why it is not considered practical to incorporate functions (such as BECN and FECN) that are unique to frame relay. However, network elements such as routers are often equipped with functions for interpreting BECN and FECN and for taking action to reduce the load - through temporary storage in buffers, for example.
Hope this helps --
Also, yes IOS 12.2 handles traffic shaping - not 100% on IOS 12.1 (unless someone else knows).
I'm actually running IOS 12.2
Hope this helps --
All of these have to do with Frame-relay traffic shaping , This is a very good page that describes Friame-relay traffic shaping
I also tried to turn on traffic-shaping for just one sub-interface but it is not possible. I guess I'll have to turn it off on the main interface and we will have to do this on all routers or no routers.
Thanks for all of the input and feedback here everyone. You guys are life savers...
Can it be enabled on just a DLCI and not the main interface?
Secondly, I was told that all routers would have to be running at IOS level 12.2 to use the traffic shaping features. All of the other routers are at 12.1 so we are not sure if we do need to upgrade or not.
You can create several class maps for various bandwidth sized circuits
I used T1, 384k & 56k and it worked fine. You just need to remember to assign every DLCI off the physical interface to a class, or they will only get 56K.
I actually found the bandwidth limitation on the default class to be 200k when using a 10000 series router w/ DS-3 card. Not sure why?
description DS-3 Sho-Me/Verizon Customer Agg. Circuit
8e3ds3-1 card, 8 ports