Re: ICMP Query!!!

OK.... is ICMP traffic marked with DSCP cs6 when it leaves the router?
 That becomes the simple answer. If it can't be part of the pak
priority queue then it is not control plane traffic

Regards,

Tyson Scott
CCIE # 13513 (R&S, Security, SP)
Managing Partner/Technical Instructor - IPexpert Inc.
tscott_at_ipexpert.com

----- Reply message -----
From: "Carlos G Mendioroz" <tron_at_huapi.ba.ar>
Date: Sat, Nov 13, 2010 3:44 pm
Subject: ICMP Query!!!
To: "Tyson Scott" <tyson.scott_at_advtechracks.com>
Cc: "negron.paul_at_gmail.com" <negron.paul_at_gmail.com>,
"ron.wilkerson_at_gmail.com" <ron.wilkerson_at_gmail.com>,
"eliteccie_at_gmail.com" <eliteccie_at_gmail.com>, "ccielab_at_groupstudy.com"
<ccielab_at_groupstudy.com>

As Marko said, moot point:
  An issue that is subject to, or open for discussion or debate;
  originally, one to be definitively determined by an assembly of the
  people.

May be the issue is the lack of a concrete definition of what is a
data plane protocol. We all know what data plane is, control plane too.
But what is a control plane protocol ? And even more confusing,
what would a data plane protocol be ?

How one router treats the traffic seems not to be a good tie breaker,
cause e.g. a multihop BGP session passes through an intermediate router
as data, and that will not make BGP a data plane protocol, would it ?

So I like the view that if a router has to pay attention, it is control
plane. Attention to the packet content that is. So that rules
out the process switching argument.
The name means a lot too, to me at least.

In any case, I don't really see a point in defining a protocol as
data plane protocol. If anyone can come up with a good definition
of that, then we can resume the discussion of ICMP being one such animal
or not...

-Carlos

Tyson Scott @ 13/11/2010 17:24 -0300 dixit:
> ICMP is not control plane traffic. ICMP unreachables go to the CEF
> exception for example. Consider the control plane as protocols that
> glue the network together. ICMP traffic to the router go to the host
> control plane because of being directed to the device thus it must
> handle it. ICMP is data traffic that may be used for management
> purposes
>
> Regards,
>
> Tyson Scott
> CCIE # 13513 (R&amp;S, Security, SP)
> Managing Partner/Technical Instructor - IPexpert Inc.
> tscott_at_ipexpert.com
>
>
> ----- Reply message -----
> From: "Paul Negron" <negron.paul_at_gmail.com>
> Date: Sat, Nov 13, 2010 2:10 pm
> Subject: ICMP Query!!!
> To: "ron.wilkerson_at_gmail.com" <ron.wilkerson_at_gmail.com>, "Carlos G
> Mendioroz" <tron_at_huapi.ba.ar>
> Cc: "CCIE KID" <eliteccie_at_gmail.com>, "Cisco certification"
> <ccielab_at_groupstudy.com>
>
>
> Very Interesting Response.
>
> I guess I primarily viewed ICMP as testing the Control Plane/ Data Plane
> with the Majority of ICMP Query types:
>
> * 0 = Echo Reply (3ping response2)
> * 8 = Echo Request (3ping query2)
> * 9 = Router Advertisement (RFC 1256)
> * 10 = Router Solicitation (RFC 1256)
> * 13 = Time Stamp Request
> * 14 = Time Stamp Reply
> * 17 = Address Mask Request
> * 18 = Address Mask Reply
>
> I know my definition is a little Narrow but it does help differentiate ICMP
> from protocols like RSVP, PIM, EIGRP that strictly represent Control Plane
> from a Routing Switching perspective.
>
> As far as the view that because ICMP uses the CPU being a CLEAR definition,
> this I would disagree with. What would Process Switching be then? Control
> Plane or Data Plane activity?
>
> Carlos and Ron do make a good point to expand my Narrow definition though.
> :-)
>
> Paul
>
>

-- 
Carlos G Mendioroz  <tron_at_huapi.ba.ar>  LW7 EQI  Argentina
Blogs and organic groups at http://www.ccie.net