RE: [Diffserv-interest] packet classification and policing

Anoop,
  You have a good point, but then it is obvious then that you cannot always
guarantee the conforming rate for both policers when microflows are running
above the conforming rate.  The two policiers pretty much should be considered
independent from each other.  Much like a Layer 3 policer wouldn't influence
a Layer 2 policer.

The order of the policers can also be determined by the SLA, you might wish
to re-order the policer, where the BA policer is first, then the microflow
policer.

I'm not aware of any implementations out there where one policer can
influence the credit system of another policer, maybe someone out there
is aware of something like that and can correct me.

--Jeff

In message "[Diffserv-interest] packet classification and policing", Anoop Ghanwani writes:

>
>Jeff, 
>
>Thanks for response. 
>
>If a packet can go through multiple policers, then 
>there is the following issue.  Suppose the 
>packet goes through two policers.  One of them says it's 
>conforming, the other says that it's out of profile.  I 
>assume then, that the packet would be treated as an 
>out-of-profile packet.  However, it has just used some 
>some "good" credits from the policer that considered it 
>conforming.  Those are wasted credits and could well 
>have been used by some other packet. 
>
>How do you think that should be handled? 
>
>-Anoop 
>
>> -----Original Message----- 
>> From: Jeff Norman [ mailto:jnorman@nortelnetworks.com
><mailto:jnorman@nortelnetworks.com> ] 
>> Sent: Thursday, August 16, 2001 1:48 PM 
>> To: Anoop Ghanwani 
>> Cc: 'diffserv-interest@ietf.org' 
>> Subject: re:[Diffserv-interest] packet classification and policing 
>> 
>> 
>> Anoop, 
>>   Although not part of a standard, I don't think there is anything 
>> preventing anyone from allowing packets from going through more 
>> than one policer.   Actually the example you gave is the most 
>> common example known IMHO of using two policiers for IP packets, 
>> one to control the microflow feeding conforming packets into 
>> a BA policer. 
>> 
>> --Jeff 
>> 
>> In message "[Diffserv-interest] packet classification and 
>> policing", Anoop Ghanwani writes: 
>> 
>> > 
>> > 
>> >When setting up packet classifiers and policers, is it 
>> >valid to assume that a packet can hit more than one policer?  
>> > 
>> >For instance, a packet which gets classified as being part 
>> >of a microflow may have a policer associated with it.  At 
>> >the same time, it may be desirable to police the aggregate 
>> >traffic from the interface that this packet came in on. 
>> > 
>> >Does the above scenario make sense, or is it expected that 
>> >a packet will hit at most one policer on ingress? 
>> > 
>> >Thanks, 
>> >-Anoop 
>> 
>########################################################################
>############################## This email communication may contain
>CONFIDENTIAL INFORMATION and is intended only for the use of the
>intended recipients identified above.  If you are not the intended
>recipient of this communication, you must not use, disclose, distribute,
>copy or print this email. If you have received this communication in
>error, please immediately notify the sender by reply email, delete the
>communication and destroy all copies.
>########################################################################
>##############################
>
>
>  [Part 2,  Text/html  4.4KB]
>  [Not Shown. Use the attachment viewer to view this part]
>

_______________________________________________
Diffserv-interest mailing list
Diffserv-interest@ietf.org
http://www1.ietf.org/mailman/listinfo/diffserv-interest