The MPLS WG Archive

[Dave McDysan <dave.mcdysan@wcom.com>]

I wouldn't want to plug any particular probe vendor, but several provide
this type of capability. Do a search on the web.

Also, generating packets (e.g., with MPLS labels or an IP address for an FEC
that maps to a LSP) is not all that hard to do in software on commerically
available hosts.

If the root cause of the problem is vendor hardware/software that is
claiming that an LSP is up but is not forwarding labeled packets, then would
you really trust that same vendor to reliably generate and detect the
absence of a heartbeat?

Dave

> -----Original Message-----
> From: owner-mpls@UU.NET [mailto:owner-mpls@UU.NET]On Behalf Of Giles
> Heron
> Sent: Friday, December 14, 2001 5:21 AM
> To: Don Fedyk
> Cc: Punj, Arun; Ben Mack-Crane; neil.2.harrison@bt.com; mpls@UU.NET
> Subject: Re: MPLSOAM BOF meeting draft minutes
>
>
> > Don Fedyk wrote:
> >
> > Giles
> >
> > What are these other ways? External equipment? That would
> > work and I am sure someone could build boxes to do that.
>
> yes.
>
> though when I said "I sell them", I meant PWs - not the boxes :)
>
> > Could you elaborate or offer a pointer?
>
> As I said in an earlier post, with OAM-deficient L2s you probably want
> some kind of external device (or a mechanism inside your edge box for
> injecting traffic "outside" the PW edge function.)
>
> With OAM-enabled L2s then you can just use the native OAM and treat the
> PW as a section of your path.
>
> Giles
>
> > Don
> >
> > > -----Original Message-----
> > > From: Giles Heron [mailto:giles@packetexchange.net]
> > > Sent: Thursday, December 13, 2001 10:58 AM
> > > To: Punj, Arun
> > > Cc: Fedyk, Don [BL60:1A00:EXCH]; Ben Mack-Crane;
> > > neil.2.harrison@bt.com;
> > > mpls@UU.NET
> > > Subject: Re: MPLSOAM BOF meeting draft minutes
> > >
> > >
> > > there are other ways to connectivity check pseudo-wires.
> > >
> > > I should know.  I sell them :)
> > >
> > > "Punj, Arun" wrote:
> > > >
> > > > Folks,
> > > >
> > > > OAM is required to ensure that customer is given a level of
> > > service that he
> > > > has contracted for.
> > > > If all you have offered is best effort delivery than IP
> > > ping is sufficient.
> > > > But if customer has
> > > > contracted for a pseudo-wire service than you better
> > > perform connectivity
> > > > checks on this pseudo
> > > > wire, or you will soon have claims piling on your desk.
> > > >
> > > > > -----Original Message-----
> > > > > From: Giles Heron [mailto:giles@packetexchange.net]
> > > > > Sent: Thursday, December 13, 2001 5:48 AM
> > > > > To: Don Fedyk
> > > > > Cc: Ben Mack-Crane; neil.2.harrison@bt.com; mpls@UU.NET
> > > > > Subject: Re: MPLSOAM BOF meeting draft minutes
> > > > >
> > > > >
> > > > > > Don Fedyk wrote:
> > > > > >
> > > > > > Giles
> > > > > >
> > > > > > Semi broken boxes was the reason that LSP Ping was introduced.
> >
> > > > >
> > > > > Agreed.
> > > > >
> > > > > But why was MPLS OAM introduced?  If we are only addressing
> > > > > the problem
> > > > > of semi-broken boxes then do we need to be constantly
> > > monitoring all
> > > > > LSPs?
> > > >
> > > > The fact of the matter is that boxes/protocols/network
> > > designs - are all
> > > > prone to errors.
> > > > The best you can do is to ensure that a problem in the
> > > network does not
> > > > result in a hit
> > > > to the services offered to the customer.
> > > >
> > > > Another point - you need not monitor (or enable CV) on all
> > > the LSPs.  You
> > > > just need to
> > > > monitor *important* LSPs.
> > > >
> > > > >
> > > > > Again, I would compare this to IP networks.  In the IP
> > > case we don't
> > > > > send pings constantly to every prefix in our networks to
> > > > > check that they
> > > > > are still reachable, so I'm not 100% convinced we should be
> > > > > sending CVs
> > > > > constantly in MPLS?
> > > >
> > > > Ask yourself this simple question:
> > > >
> > > > Would you rather like your customers to find bugs in your
> > > network or are you
> > > > better
> > > > off finding the bugs yourself? OAM CV lets you find bugs before
> > your
> > > > customers find them.
> > > > Presumably there might be a few customers in your network
> > > who do not mind
> > > > debugging your
> > > > network for you. But surely the ones contracting for
> > > premium services like
> > > > VPNs/PWs would not
> > > > be thrilled to do so...
> > > >
> > > > BTW, more than one network operator felt that finding bugs
> > > in the networks
> > > > before customer do
> > > > so is important to them. Maybe their customers were more
> > > difficult kind...
> > > > May be they were just
> > > > being irrational... Maybe these operators really do know
> > > what is right.
> > > >
> > > > As for not pinging all the IP prefixes in this world
> > > comtinuously... Once
> > > > you implement
> > > > pseudo-wires or VPN type applications over IP I have a
> > > feeling you might
> > > > think that some
> > > > IP prefixes are worth pinging all the time. ( although even
> > > in that case,
> > > > ping should be
> > > > replaced with CV).
> > > >
> > > > Arun
> > > >
> > > > --Snipped---
> > >
> > > --
> > > =================================================================
> > > Giles Heron    Principal Network Architect    PacketExchange Ltd.
> > > ph: +44 7880 506185              "if you build it they will yawn"
> > > =================================================================
> > >
>
> --
> =================================================================
> Giles Heron    Principal Network Architect    PacketExchange Ltd.
> ph: +44 7880 506185              "if you build it they will yawn"
> =================================================================