The MPLS WG Archive

[Marc Binderberger <marc@sniff.de>]

Hi Sylvia,

> May I be bold enough to ask the reason MPLS was ever
> deployed?

good marketing opportunity? ;-)

Serious: there are probably some folks on the list who invented MPLS, 
being able to give you all reasons. Here a few from my limited point of 
view:

* the speed argument: MPLS is now 8 years old (I think. or even 
older?), at that time Silicon to swap labels had been available for ATM 
already while hardware-based routing was still expensive, less mature 
and so on. _Today_ the argument is not valid anymore - but that's 8 
years later.

* Larger ISPs in the US often had an ATM core with routers at the edge 
and any-any PVC meshing. What is called the overlay model. Doesn't 
scale well for your IGP, so making the ATM switches part of your 
IGP/MPLS (the "peer" model) scales better. For invest protection etc. 
you want to use the existing ATM boxes.

Then we had the Internet bubble. Bandwidth for free (well, or not). 
Suddenly you had 10Gbit/s networks out there - and a need to use them.

* Do you wanna build a separate ATM/FR network? No, you want similar 
capabilities on your router network, so you get TE.

* Fast rerouting because these new networks often are DWDM, not SDH 
based anymore (read: unprotected wavelength).

Additionally some bright minds realized how cool the routing protocols 
are in the IP world. Instead of SDH rings - somewhat limited - it would 
be more flexible and even cheaper (hmm) when you run meshed networks. 
So suddenly you have stuff like OSPF on your transmission equipment for 
circuit setup etc ... well, you need more than OSPF and (G)MPLS as the 
generic control plane comes into play here.


As I said: just a few reasons. Many different motivations meanwhile.


> And how many such end addresses are there? Do we need
> to throw out the existing gear to move to IPv6?

this problem is not related to MPLS, is it? In fact I know hardware 
where MPLS is the only way to run IPv6 at high speed - because the MPLS 
"hides" the IPv6 which some ASICs cannot deal with (and you don't wanna 
run STM16 on a 200MHz MIPS R5000 ;-)

> But then tomorrow's ASICs will be faster than today's
> so why not wait for tomorrow :)) till they are built
> ??

faster than wire speed? ;-)

> But why woould I need ATM then?

See above: some _have_ ATM networks and want to integrate IP and ATM 
networks into one.


Regards, Marc

P.S.: is this "historical" discussion Off-Topic for the list? In this 
case I apologize and switch over to private comm channels :-)
--
Marc Binderberger    <marc@sniff.de>


_______________________________________________
mpls mailing list
mpls@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/mpls