The MPLS WG Archive

[Shahram Davari <Shahram_Davari@pmc-sierra.com>]

> -----Original Message-----
> From: Thomas D. Nadeau [mailto:tnadeau@cisco.com]
> Sent: Thursday, March 21, 2002 9:35 AM
> To: Shahram Davari
> Cc: 'mpls@uu.net'
> Subject: RE: Y.1711 requirements from IETF
> 
> 
> At 03:31 PM 3/20/2002 -0800, Shahram Davari wrote:
> >Thomas,
> >
> >
> > > > >
> > > > >          You should be clear. It specifies that *all* 
> LSPs must be
> > > > > provisioned with this feature
> > > > > on and at a specific interval.
> > > >
> > > >
> > > >Where did you get this " ALL LSPs MUST" statement from? May
> > > be you are
> > > >reading a different Y.1711 than I have. All I can find are
> > > the following
> > > >statement:
> > > >
> > > >" ... It is strongly recommended that CV OAM packets be
> > > generated on all LSPs
> > > >(in order to detect all defects and potentially provide
> > > protection against
> > > >traffic leakage both in and out of LSPs). "
> > > >
> > > >"... CV generation could be enabled on all (or most) of the LSPs"
> > > >
> > > >There is no MUST and in fact you could enable the CV 
> only on ONE LSP.
> > > >If the text is not conveying this information, then we would have
> > > >to add additional text or appendix in the next version 
> to clarify it.
> > >
> > >          The text at strongly implies that it should be run
> > > everywhere all
> > > the time.
> > > Not knowing the ITU language mapping from IETF MUST, MAY or
> > > SHOULD, it is a reasonable assumption that the text you refer to
> > > requires OAM to be run on every LSR and on every LSP.
> > >
> >
> >I think you are imagining things. Now you added one more 
> imagination to your
> >comments : "..requires OAM to be run on every LSR". Where 
> did you get this one
> >from?
> >
> >For clarification OAM needs to be run only on edge LSRs, and 
> you could
> >turn it on or off on any LSP that you want.
> 
>          Two points. First, rarely is an LSR just an LSR, the argument
> that this  needs to run "only on edge LSRs" could in reality mean
> that it needs to run on every or nearly every LSR in a 
> provider network.
> Second, given the fact that many LSRs that also have PE 
> functionality also
> contain the entire Internet routing database when running LDP for LSP
> tunnels, running this on devices doing edge duty at the "strongly
> recommended" rate of 1 ping per LSP has ramifications for 
> network bandwidth
> on potentially every device.  Again, I recommend scaling back your
> "strongly recommended" terminology, as it is clear that this is not
> a solution that fits all networks.

But the same argument applies to MPLS-ping too. You also need to upgrade all LSRs.

-Shahram





> 
>          --Tom
> 
> 
> 
> 
> 
> --------------------------------------------------------------
> ----------
> Mathematics is the supreme nostalgia of our time. 
>