Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 29 May 2002 11:22:39 -0400
From:      "Louis A. Mamakos" <louie@TransSys.COM>
To:        Andre Oppermann <oppermann@pipeline.ch>
Cc:        Attila Nagy <bra@fsn.hu>, Luigi Iannone <Luigi.Iannone@lip6.fr>, freebsd-net@FreeBSD.ORG
Subject:   Re: MPLS 
Message-ID:  <200205291522.g4TFMdRG076033@whizzo.transsys.com>
In-Reply-To: Your message of "Wed, 29 May 2002 16:24:03 %2B0200." <3CF4E483.2510639@pipeline.ch> 
References:  <Pine.NEB.4.44.0205290915060.28431-100000@tibre.lip6.fr> <Pine.LNX.4.44.0205291108080.7798-100000@scribble.fsn.hu> <3CF4A64A.EE220611@pipeline.ch> <200205291413.g4TEDLRG075458@whizzo.transsys.com> <3CF4E483.2510639@pipeline.ch> 

next in thread | previous in thread | raw e-mail | index | archive | help

> If there is no kind of software involved on the forwarding plane
> then I don't know how the control plane can communicate via ethernet
> with the line cards... The internal communication in the router is
> via ethernet.

To be clear, "forwarding plane" to me means the machinary which
causes packets that are received on one network interface on the
router to be routed and sent out another (or same) interface on
the router.  "Control plane" is the stuff which executes the
routing protocols, and does various management tasks in the
router, including setting up forwarding tables used by the ASIC
hardware. 

The control plane (e.g., the route processor) is connected to
the forwarding infrastructure so it's able to send and receive
traffic via the other ports on the box.  

Do not confuse the ethernet on the control processor with the
"real" (Gigabit) Ethernet interfaces which plug into the line cards.
They source and sink traffic just like the various other line card
interfaces (various POS and ATM interfaces.)

Also, there's no fowarding table on the line cards, either.  The
ASIC (one on the M5, M10, M20 and M40, 4 on the M160) are associated
with the backplane of the router.

Further detail may be covered by non-disclosure agreements which
I'm subject to.

> > The forwarding is done in Juniper's custom designed ASIC hardware,
> > and is the other significantly valuable intellectual property
> > they have along with the routing protocol implementation (e.g.,
> > BGP, IS-IS, etc.)
> 
> I agree with the ASIC hardware. But the BGP implementation smells
> awfully like gated (Nexthop). Anyway, a BGP deamon isn't that hard
> to write.

Please don't take this the wrong way, but this sounds like a statement
from someone that's never done it before.  In my professional role,
I've seen literally a dozen or more vendors get the routing protocol
implementation Really Wrong.  Sure, building and parsing BGP protocol
packet isn't that hard; the interesting bit is the policy machinary
which filters and computes routes.  For a router in the default-free
core of the Internet, this is a rather daunting task.  While the
JunOS implementation might smell like gated, it has significantly
more functionality and reliability than /usr/ports/net/gated does.

louie


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200205291522.g4TFMdRG076033>