Frame Relay Traffic Shaping Summary

GTS

- Interface/subinterface level (no pvc level)
- Internal shaper is WFQ
- Could be used with WFQ/PQ/CQ at interface level

interface Serial0/0/0:0.1 point-to-point
ip address 177.0.112.1 255.255.255.0
traffic-shape rate 512000
traffic-shape adaptive 256000
traffic-shape fecn-adapt
frame-relay interface-dlci 112


Legacy FRTS

- Enabled with frame-relay traffic-shaping
- PVC granularity
- Default to 56k/125ms once activated
- Could be used with WFQ/PQ/CQ per PVC
- Could not be used with WFQ/PQ/CQ at interface level (if FRF.12 2 dual fifo is enabled at interface level)
- Could use FRF12 of the map-class, per VC fragmentation

map-class frame-relay SHAPE_384K
frame-relay cir 384000
frame-relay bc 3840
frame-relay be 0
frame-relay adaptive-shaping becn
frame-relay adaptive-shaping interface-congestion
frame-relay fair-queue

interface Serial 0/0/0:0
frame-relay traffic-shaping
!
interface Serial 0/0/0:0.1
ip address 177.0.112.1 255.255.255.0
frame-relay interface-dlci 112
class SHAPE_384K



MQC based FRTS

- Needed for Voice-Adaptive Traffic Shaping
- Could use FRF 12 at interface level (for all vc of the interface)
- Incompatible with the command frame-relay traffic-shaping

policy-map CBWFQ
class VOICE
priority 64
class DATA
bandwidth 128
class class-default
fair-queue

policy-map SHAPE_384K
class class-default
shape average 384000
shape adaptive 192000
service-policy CBWFQ

map-class frame-relay SHAPE_384K
service-policy output SHAPE_384K

interface Serial 0/0/0:0.1
ip address 177.0.112.1 255.255.255.0
frame-relay interface-dlci 112
class SHAPE_384K


MQC based Class based traffic shaping

- Use map-class
- Incompatible with frame-relay traffic-shaping
- Incompatible with adaptive shaping
- Use service policy with shaping configured on default-class or by pvc matching a dlci
- CBFWQ could be used to child policy
- FRF.12 is applied at interface level

policy-map CBWFQ
class VOICE
priority 64
class DATA
bandwidth 128
class class-default
fair-queue

policy-map SHAPE_384K
class class-default
shape average 384000
shape adaptive 192000
service-policy CBWFQ

interface Serial 0/0/0:0.1
ip address 177.0.112.1 255.255.255.0
service-policy output SHAPE_384K
frame-relay interface-dlci 112

3550 QOS : WRR

On 3550 : Weighted Round Robin is done per packet
On 6000/6500 : Weighted Round Robin is done by bandwidth

wrr-queue bandwidth 10 20 30 40

On gigabit ports:

The 4 queues could be assigned a different amount of buffer size and WRED or RED with 2 thershold by queues :

Wred :
wrr-queue random-detect max-threshold 1 50 100

Red :
wrr-queue threshold 1 50 100

Queue Size:
wrr-queue queue-limit Q_size1 Q_size2 Q_size3 Q_size4

On non-gigabit ports:

Only FiFO with a reserved number of packet by queues, (max 170) :

mls qos min-reserve level buffer_size
wrr-queue min-reserve level queue

3560 QOS

Differents tasks that could be asked :

1 Map Ingress DSCP value to different queues and WTD :

There is 2 input queues each with 2 different WTD threshold

Followinbg example maps dspc 00 and 01 to the queue 1 threshold 2 and configure for queue 1 and threshold 1 of 75 and threshold 2 of 50 and for queue 2 a threshold1 of 30 and a threshold 2 of 75

mls qos srr-queue input dscp-map queue 1 threshold 2 00 01
mls qos srr-queue input threshold 1 75 50
mls qos srr-queue input threshold 2 30 75


2 Configure the ingress bandwidth association to queues

Here 20 % is dedicated to Expedite queue 2, the remain bandwidth is shared between queue 1 35% and queue 2 45%

mls qos srr-queue input bandwidth 35 45
mls qos srr-queeu input priority 2 bandwidth 20

3 Modifiy default buffer

Buffer is the amount of packet each queue could contains before dropping

Input buffer is globally modified with :
mls qos srr-queue input buffer 60 40

Ouput is modified by Queue-set. Queue-set 1 is by default applied to all interfaces with egal shared. To apply a different queue-set to an interface :

mls qos queue-set output 2 buffers 40 20 20 20
int fa0/14
queue-set 2

4 Configure bandwidth shaping and sharing on interface

Shaped queue limit the bandwidth, whereas shared only limit during congestion.
Shape weight is an inverse ration 1/weight
Share weight is a ration weight_queue1/total_weight
0 in shape means queue is treated as shared. The remain bandwith is for shared.
A queue in shape mode is not take into account on the shared ratio.

Example Queue 1 is a shared queue with 12,5 percent of the bandwidh, the remaining bandwidth is shared among q2, q3, q4.

srr-queue bandwidth shape 8 0 0 0
srr-queue bandwidth share 25 25 25 25

Priority Queueing

In priority queueing, packets are assigned to one of the 4 availables queues according to protocol type, acl or entering interface. The differents queues are served in a strict priority fashion, that is when a packet has to been sent, queue 1 is always served before until limit or emptied. It could lead to bandwidth starvation (onlys queue 1 is served)

Configuration

Assign the Priority-list to an interface
interface fa1/0
priority-group list-num

Assign packet to a queue
priority-list list-num protocol protocol high|medium|normal|low list|tcp|udp list-or-proto-num
OR
priority-list list-num interface int-typ int-num high|medium|normal|low

Configure queue length
priority-list list-num limit high-limit medium-limit normal-limit low-limit

Custom Queuing

It permits to share bandwidth accross up to 16 different queues according to protocol, acl, entering interface. The sharing is done by maximum packet count by queues or average bytes count by queues.

Configuration :

Apply on interface
interface fa0/1
custom-queue-list list-num

Define queue size
custom-queue-list list-num queue queue-num limit packet-limit
OR
custom-queue-list list-num queue queue-num byte-count byte-limit

Assign packet to queues
queue-list list-number protocol protocol-name queue-number list|tcp|udp keyword-value
OR
queue-list list-number interface int-type int-num queue-number

Traffic Policing : Rate-Limit or MQC Policing

Example : limiting icmp traffic to 128kbps with a permitted burst during 1/4th of the rate

Legacy traffic policing using the command rate-limit

access-list 110 permit icmp any any
access-list 110 permit ip any any

interface e0/0
rate-limit 110 128000 4000 4000 conform-action transmit exceed-action drop

128000 bps : rate-limit
4000 bytes : normal rate that could be sent at clock rate (bucket size every timeslot)
4000 bytes : exceed rate that could be sent at clock rate

Bc = Be so there is no exceed burst

Verification

show interface rate-limit


traffic policing using MQC

Cef must be activated

class-map match-all CAR
match protocol icmp

policy-map CAR
class CAR
police cir 128000 bc 4000

interface e0/0
service-policy out CAR

128000 : rate limit
4000 : normal rate that could be sent at clock rate
Be : 0

Verification

show policy-map interface


Differences

- Legacy CAR : configured Bc should include Be (Bc = excess_bits + Be)
- MQC CAR use a default Bc of 2xBe if not configured
- MQC CAR : configured Bc doesn't include Be ( Bc = excess_bits)