clarification wanted In Class Base Shapping

vinay_verma80 · ‎08-08-2007

1 what is the CB shapping defaults values more specifically where ?=? belong ( rate <=320kbps or rate >= 320kbps)

as in 2 books I find different version( I hope u don?t have to get the router and perform the test)

In cisco QOS exam cert Guide by Wendell Odom(page 371 )

Variable ------------ rate <320kbps --------------- rate >= 320kbps

Bc ------------------ 8000bits ---------------------------- Bc=CIR* Tc

Be------------------------ Bc=Be=8000bits----------------------Be=Be

Tc------------------------- Tc=Bc/CIR----------------------------25ms

But in CCIE Offical exam cert guide by same author Wendell Odom

Variable-------------------rate <=320kbps-------------------------rate > 320kbps

Bc-------------------------- 8000bits--------------------------------Bc=CIR* Tc

Be---------------------------Bc=Be=8000bits-----------------------Be=Be

Tc------------------------------Tc=Bc/CIR----------------------------25ms

2. we know that

in shapping average with exess burst the command is

shape average 64000

where be= be

(a) If we have to shape with out excess burst does the command is like

shape average 64000 bc 0

where bc= valued specified

be=0

or

we can write

shape average 64000 bc

where bc= valued specified

be = ?? >>>>>>>>>>>>>>>>>> it will take by default 0 ( as only bc specified) or still be=bc or something else

also is there is any restriction with the bc value according to formala (Bc=CIR* Tc)

3. I know that we can specify queing on physical interface ( WFQ , PQ , FIFO) directly with out CB queing

If I apply WFQ /FIFO on a phicical interface which is having 2 subinterrface

And the we have applied CB Shapping on subinterface which ?have? nested LLQ CB shapping( ie LLQ queu is netsed into shapping) how will it be going to main interface WFQ /FIFO ( if so what is then the use of llq as every thing will still will pass WFQ/FIFO from amin queue)

Thanks in advance

Regards

Vinay verma rolleyes.gif

Pavel Bykov · ‎08-09-2007

Just quickie for Number1:

Actually, both formulas are right for 320. That's where the values meet. So it doesn't really matter. 320/8 is 40. A nd 1/40 is 0,025s, which is 25ms. So if you use a formula you get 25ms, and if you use static TC, you get the same result. So it depends on implementation, but does not make a difference either way.

vinay_verma80 · ‎08-10-2007

thanks

can u help me on the step 3 also

regards

Pavel Bykov · ‎08-11-2007

For no.3:

All queues are FIFO. (except special cases like LIFO). With queueing strategies, you are just creating more FIFO queues. Even WFQ creates many fifo queues - one for each flow.

Actually you can find an answer to this in practice. In theory, Wendell does say that subinterface queue is then dispatched into interface queue.

BUT

1: many times it is not allowed by router. i.e.:

2851_hra_lab(config)#int Serial0/0/0:0.1 point-to-point

2851_hra_lab(config-subif)#service-policy output PM-LLQ_FAST-OUT

CBWFQ : Not supported on subinterfaces

or it there can be some restrictions:

2851_hra_lab(config-if)#fair-queue

2851_hra_lab(config-if)#service-policy output PM-LLQ_FAST-OUT

Must remove fair-queue configuration first.

But in a lot of cases, what IOS does is it suspends queueing strategy. Check out the show interface command output:

Queueing strategy: weighted fair [suspended, using FIFO]

While it says that it "uses FIFO", in reality the LLQ policy output of my map on subinterface goes directly into tx_ring.

Here is an example of this:

http://www.cisco.com/warp/public/788/voice-qos/voip-mlppp.html

And here is a nice proof:

It's serial with Multilink PPP. MPP has LLQ output map. It's 1/4 of E1 link and is 512 kbps. I loaded it with 1024 kbps traffic (50% drop rate). Here is a show interface output:

Serial0/0/1:0 is up, line protocol is up

Hardware is GT96K Serial

Description: sp_7206_hra_lab

MTU 1500 bytes, BW 512 Kbit, DLY 20000 usec,

reliability 255/255, txload 252/255, rxload 253/255

Encapsulation PPP, LCP Open, multilink Open

Link is a member of Multilink bundle Multilink2363, loopback not set

Keepalive set (10 sec)

Last input 00:00:00, output 00:00:00, output hang never

Last clearing of "show interface" counters 3d01h

Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 0

Queueing strategy: weighted fair [suspended, using FIFO]

FIFO output queue 0/40, 0 drops

30 second input rate 508000 bits/sec, 130 packets/sec

30 second output rate 507000 bits/sec, 130 packets/sec

1305092 packets input, 220812748 bytes, 0 no buffer

Received 0 broadcasts, 0 runts, 0 giants, 0 throttles

0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored, 0 abort

1842975 packets output, 573165503 bytes, 0 underruns

0 output errors, 0 collisions, 1 interface resets

0 output buffer failures, 0 output buffers swapped out

2 carrier transitions

Timeslot(s) Used:1-8, SCC: 1, Transmitter delay is 0 flags

Notice the line:

FIFO output queue 0/40, 0 drops

2. In case of Frame relay and FRTS there are frame-relay maps. Then there is a second output queue. But in case of FRF.12, there is a DUAL-FIFO mechanism. If there would be normal fifo, it would work as an extension to tx_ring (one fifo going into the other). But DUAL-FIFO actually diferentiates between LLQ and CBWFQ classes and places the packets accordingly.

Hope this helps for number 3.