From: Mike Williams (ccie2be@swbell.net)
Date: Fri Nov 21 2003 - 16:41:34 GMT-3
The main difference between shaping and policing in that situation is
that policing will drop and shaping will at least buffer some of the
traffic to smooth out the peaks. One side effect of this, for instance,
is that policing will cause TCP retransmits (or cause dropped UDP
traffic to be totally lost) whereas shaping will allow more of the
traffic through, just delayed a bit (not always the best for
time-sensitive stuff, but like with streaming audio/video, as long as
the delay is smaller than the buffer on the receiving station, there
shouldn't be a problem). However, using shaping does not ensure that
none of the traffic will be dropped, it simply gives bursty traffic a
much better change of getting through compared to policing.
This document has a good graphical representation of what shaping and
policing do:
http://www.cisco.com/warp/public/105/policevsshape.pdf
Mike W.
-----Original Message-----
From: Anthony Pace [mailto:anthonypace@fastmail.fm]
Sent: Friday, November 21, 2003 9:30 AM
To: Mike Williams; 'asadovnikov'; 'Hunt Lee'; ccielab@groupstudy.com
Subject: RE: Which one should I pick ?
If shaping is ont during times of non-congestion then how does it differ
from policing in the scenario below? Why ot just polic it if you want a
strict cap on bandwidth?
Tony
On Wed, 8 Oct 2003 17:15:49 -0500, "Mike Williams" <ccie2be@swbell.net>
said:
> Not neccesarily true... If you use GTS on an interface, then (AFAIK)
> it will only kick in when there is congestion, but if you apply
> shaping using a policy map (i.e. class-based shaping), the shaping
> "creates a false congestion" whenever the that traffic class exceeds
> the values you use on the shaping command (this was told to me by a
> group of Cisco
> SE's) and then actively shapes the traffic. I configured the policy
to
> shape certain traffic down to 8Kbps, and it works like a charm
> regardless of whether the interface was congested or not.
Furthermore,
> my testing was also done on an ethernet interface, and this was a 3600
> router (not really "high-end"). In my testing, we were trying to
limit
> the bandwidth used by a VPN tunnel. Since we knew the IP address of
> each endpoint, I just made an access-list, made a class that matched
> that ACL, then made a policy that only affected that class. Since you
> can only shape on traffic going out of an interface, I had to apply
this
> shaping to the FastEthernet interface (instead of the ATM interface).
> Like I said, it worked like a charm, and I could see that it was
> actively shaping the traffic down to 8Kbps, as well as how many
packets
> were delayed, dropped, etc...
>
> Mike W.
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf
> Of asadovnikov
> Sent: Wednesday, October 08, 2003 5:51 AM
> To: 'Hunt Lee'; ccielab@groupstudy.com
> Subject: RE: Which one should I pick ?
>
>
> The way you put the question makes policing the only choice. Either a
> police statement as you specified, or older CAR will do. Shaping only
> kicks in during congestion periods and is not in effect unless a
> congestion is present.
>
> Further I do not believe shaping is supported on Ethernet (some higher
> end boxes to have additional QOS implemented in hardware).
>
> Best regards,
> Alexei
>
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf
> Of Hunt Lee
> Sent: Tuesday, October 07, 2003 10:06 PM
> To: ccielab@groupstudy.com
> Subject: Which one should I pick ?
>
>
> If I have a question:
>
> Configure CBWFQ on R1's Ethernet interface such that it allocates min
> bandwidth of 3mbps for VLANA. VLANA traffic should limit to 5mbps and
> cannot utilize more physical bandwidth during non-congestion period.
>
> Should I use:
>
> policy-map cbwfq
> class VLAND
> bandwidth 3000
> shape peak 5000000
>
> OR
>
> policy-map cbwfq
> class VLAND
> bandwidth 3000
> police 5000000 937500 1875000 conform-action transmit exceed-action
> drop violate-action drop
>
> And if you could give me reasoning on why you pick that one, it would
> be greatly appreciated.
>
> Thank you so much in advance,
> H.
>
> ***Get your CCIE and a FREE vacation: Shop.GroupStudy.com***
> ______________________________________________________________________
> _
> Please help support GroupStudy by purchasing your study materials
from:
> shop.groupstudy.com
>
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
> ***Get your CCIE and a FREE vacation: Shop.GroupStudy.com***
> ______________________________________________________________________
> _
> Please help support GroupStudy by purchasing your study materials
from:
> shop.groupstudy.com
>
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
> ***Get your CCIE and a FREE vacation: Shop.GroupStudy.com***
> ______________________________________________________________________
> _
> Please help support GroupStudy by purchasing your study materials
from:
> shop.groupstudy.com
>
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
-- Anthony Pace anthonypace@fastmail.fm-- http://www.fastmail.fm - Accessible with your email software or over the web
This archive was generated by hypermail 2.1.4 : Fri Dec 12 2003 - 12:29:15 GMT-3