Re: Random-Detect Exponential vs Flow Average

Hello Eshedolonie,
Looking for information is great, but sometime is better to play with the
router, they give you great information, to understand a particular
technology, and what could be occurring

R2#conf ter
Enter configuration commands, one per line. End with CNTL/Z.
R2(config)#int s1/0
R2(config-if)#do show queueing interface s1/0 | in Exp|flow
    Exp-weight-constant: 9 (1/512)

WRED does not calcuate the drop probabluty using the current queue length,
it uses an average queue.
This queue is contantly calculated, using the previous calculated average
queue size and the current queue size
and que weighint contant is used to monitor these.

You can find at these link an exiting cisco document explaning how WRED
works with flows now
http://www.cisco.com/en/US/docs/ios/12_0t/12_0t3/feature/guide/flowwred.html

Within the CBWFQ, WRED us used to perfrom per queue dropping withing a Class
queue, each class can have its own WRED method, wicah be be weighed based on
IP Precedence or DSCP Values, so each queue can be configured with a sepatre
dropping policy

R2(config-if)#do show queueing interface s1/0
Interface Serial1/0 queueing strategy: random early detection (WRED)
    Random-detect not active on the dialer
    Exp-weight-constant: 9 (1/512)
    Mean queue depth: 0
    Max flow count: 256 Average depth factor: 16
    Flows (active/max active/max): 0/0/256

   dscp Random drop Tail drop Minimum Maximum Mark
                  pkts/bytes pkts/bytes thresh thresh prob
   af11 0/0 0/0 33 40
 1/10
   af12 0/0 0/0 28 40
 1/10
   af13 0/0 0/0 24 40
 1/10
   af21 0/0 0/0 33 40
 1/10
   af22 0/0 0/0 28 40
 1/10
   af23 0/0 0/0 24 40
 1/10
   af31 0/0 0/0 33 40
 1/10
   af32 0/0 0/0 28 40
 1/10
   af33 0/0 0/0 24 40
 1/10
   af41 0/0 0/0 33 40
 1/10
   af42 0/0 0/0 28 40
 1/10
   af43 0/0 0/0 24 40
 1/10
    cs1 0/0 0/0 22 40
 1/10
    cs2 0/0 0/0 24 40
 1/10
    cs3 0/0 0/0 26 40
 1/10
    cs4 0/0 0/0 28 40
 1/10
    cs5 0/0 0/0 31 40
 1/10
    cs6 0/0 0/0 33 40
 1/10
    cs7 0/0 0/0 35 40
 1/10
     ef 0/0 0/0 37 40
 1/10
   rsvp 0/0 0/0 37 40
 1/10
default 0/0 0/0 20 40
 1/10

Before you consider the advantages that use of flow-based WRED offers, it
helps to think about how WRED (without flow-based WRED configured) affects
different kinds of packet flows. Even before flow-based WRED classifies
packet flows, flows can be thought of as belonging to one of these
categories:
b"Nonadaptive flows, which are flows that do not respond to congestion
b"Robust flows, which on average have a uniform data rate and slow down in
response to congestion
b"Fragile flows, which, though congestion-aware, have fewer packets buffered
at a gateway than do robust flows

b"Enable WRED, using the random-detect command.
b"Optionally, configure the weight factor used in calculating the average
queue length using the random-detect exponential-weighting-constant command.
b"Optionally, configure parameters for packets with a specific IP Precedence
using the random-detect precedence command.

R2(config-if)#random-detect flow
R2(config-if)#do show queueing interface s1/0 | in Exp|flow
    Exp-weight-constant: 9 (1/512)
    Max flow count: 256 Average depth factor: 4

The random-detect flow average-depth-factor command is used to configure a
scaling factor of 4 (default) for this interface. The scaling factor is used
to scale the number of buffers available per flow and to determine the
number of packets allowed in the output queue of each active flow before the
queue is susceptible to packet drop. As the output shows, the maximum flow
count for this interface was set to 16 by the random-detect flow count
command.

R2(config-if)#random-detect flow average-depth-factor ?
  <1-16> integer

R2(config-if)#random-detect flow average-depth-factor 15
average-depth-factor must be a power of 2 (1, 2, 4, 8 or 16)

R2(config-if)#random-detect flow average-depth-factor 16
R2(config-if)#do show queueing interface s1/0 | in Exp|flow
    Exp-weight-constant: 9 (1/512)
    Max flow count: 256 Average depth factor: 16
R2(config-if)#

Here is how flow-based WRED ensures fairness among flows: it maintains a
count of the number of active flows that exist through an output interface.
Given the number of active flows and the output queue size, flow-based WRED
determines the number of buffers available per flow.
To allow for some burstiness, flow-based WRED scales the number of buffers
available per flow by a configured factor and allows each active flow to
have a certain number of packets in the output queue. This scaling factor is
common to all flows. The outcome of the scaled number of buffers becomes the
per-flow limit. When a flow exceeds the per-flow limit, the probability that
a packet from that flow will be dropped increases.

Just my 2 cents..

Victor Cappuccio
CCIE R/S# 20657
CCSI# 30452
www.anetworkerblog.com
www.linkedin.com/in/vcappuccio

2009/5/8 WW)W WWWW W <eshedalonie_at_gmail.com>

> Can someone please explain
>
> The Diffrences Between
>
> Random-detect exponential-weighthing-constant
> and
> Random-detect flow average-depth-factor ?
>
> Thanks In Advance.
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
>
>
>
>
>
>
>

--
Blogs and organic groups at http://www.ccie.net