1. Home
  2. Cisco Systems
  3. traffic shaper in class-default doesn't get all the packets

traffic shaper in class-default doesn't get all the packets

Hi there!

What is the reason, that the traffic shaper doesn't get all the packets, although the shaper is in the class class-default?

#sh policy-map interface virtual-access 11 output Virtual-Access11

Service-policy output: test-down

Class-map: class-default (match-any) 16939 packets, 16403213 bytes 30 second offered rate 391000 bps, drop rate 0 bps Match: any Traffic Shaping Target/Average Byte Sustain Excess Interval Incr. Rate Limit bits/int bits/int (ms) (bytes) 512000/256000 1984 7936 7936 31 1984

Adapt Queue Packets Bytes Packets Bytes Shaping Active Depth Delayed Delayed Active - 0 21 294 0 0 no

As you can see above, 16939 packets matched, but the shaper only saw 21 of them. As you can imagine, the wanted shaping effect is not achieved...

This virtual-access interface is derived from the following template:

interface Virtual-Template1 ip unnumbered Loopback7 ip mtu 1492 load-interval 30 peer ip address forced no peer default ip address ppp authentication chap callin ppp direction callin ppp ipcp dns 1.2.3.4 1.2.3.5

The template is for pppoe sessions over a FastEthernet (not ATM).

bba-group pppoe global virtual-template 1 sessions max limit 200 ac name FOO sessions per-mac limit 5 sessions auto cleanup

The policy-maps are quite straightforward:

policy-map test-down class class-default shape peak 256000 policy-map test-up class class-default police 256000

The service-policy for the pppoe-interface is allocated by AAA from a radius-server. By the way, policing works! It ist just the shaper, which isn't doing what it is supposed to do...

I hope, someone can help!

Thx in advance Thomas

Reply to
Thomas Fritz
Loading thread data ...

If I recall, the shaper will only become active when the interface bandwidth is exceeded. Try setting the bandwdith on the virtual- template or the physical interface to 256k. Policing does not take the Interface bandwidth into consideration at all - it's a hard limit.

If you are trying to shape to 256k use "shape average" rather than "shape peak".

Finally, the QOS implementation in some IOS versions is extremly dodgy. I was using 12.4(11T2) on a 2811 last week and the results were aweful when compared to a 1760 running 12.3(12a).

James

Reply to
James

Hi James!

Setting the bandwidth doesn't change anything.

Is it possible, that it is a queue issue (interface hardware queue, fastethernet FIFO software queue, virtual-access FIFO software queue,...)?

James schrieb:

I experienced a different behaviour: shaping to 5Mbit/s on a physical FastEthernet interface, without changing the default bandwidth, worked perfectly. I was able to see the shaping process becoming active and inactive.

#sh int fastEthernet 1/0 FastEthernet1/0 is up, line protocol is up

MTU 1500 bytes, BW 100000 Kbit, DLY 100 usec, reliability 255/255, txload 7/255, rxload 2/255 Encapsulation ARPA, loopback not set Keepalive set (10 sec) Half-duplex, 100Mb/s, 100BaseTX/FX

#sh policy-map interface fastEthernet 1/0 output FastEthernet1/0

Service-policy output: 3M-down

Class-map: class-default (match-any) 5905 packets, 5292596 bytes 30 second offered rate 707000 bps, drop rate 0 bps Match: any Traffic Shaping Target/Average Byte Sustain Excess Interval Incr. Rate Limit bits/int bits/int (ms) (bytes) 3145728/3145728 98304 393216 393216 125 49152

Adapt Queue Packets Bytes Packets Bytes Shaping Active Depth Delayed Delayed Active - 0 5905 5292596 1451 1284535 yes

It seems to me, that "Shaping Active" is set to "yes", when packets get delayed at that exact moment of time when entering the show command.

You are right. I was experimenting quite a bit...

I had the same results in this case with a 12.3(11T) and a 12.4(16).

Thomas

Reply to
Thomas Fritz

Not sure about the queuing stuff, that's beyond my knowledge.

I just noticed that you have excess burst configured though, try shape average 256000 2560 0 instead. This will shape to 256k, place data onto the wire every 10ms and disable excess burst.

I guess there could be an issue with shaping on Virtual Interfaces?

James

Reply to
James

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.