On Tue, 26 Jun 2007, Thomas Bridge wrote:
> On 25/06/07, paul at clubi.ie <paul at clubi.ie> wrote:
>>> You're saying that building layers of networks through tunneling,
>> with each layer running independent routing, is a good thing? There
> Yes. The ability for me to run my own routing protocol and policy,
> independent of my ISP, and using their backbone to connect my sites,
> is a good thing.
Right, it would be.
That's sort of beside the point though. Achieving above by doing it
in fabric of the network, rather than the control plane, by way of
tunnels is not quite as obviously good (it's pretty crap really).
>> - desire to be able to route multiple IP address spaces, which comes
>> about in large part because of the lack of public space
> You're confusing MPLS with a specific application there of. And
> the motivation behind VPNs is not as simple as the fact that some
> organisations have overlapping address ranges.
I did give a list of reasons...
Each reason /for/ MPLS boils down to a short-coming (perceived or
real, debatable or otherwise) of IP.
>> - hardware guys aversion to parsing IP (not fixed length, has options
>> that might need to be parsed, no fixed-length network versus host
>> identifiers to make hardware lookups cheaper to implement)
>> Most hardware routers can happily handle IP provided it sticks to a
> standard header length. Most of the options aren't used, and packets
> that do have them are punted to the CPU anyway.
Right, they can, of course, that's not quite the point:
An MPLS switched network is *not* parsing IP (bar ingress and
egress). Which is the point here..
>> - The administrative overheads of IPv4
>> (must configure public address per link for routing)
>> Most MPLS cores, IME, using public IP addressing on each link. So
> this is hardly a fix.
Right, except for an MPLS switching network IP is just used as an
access technology effectively - it could just as well use DECNet or
ISO without affecting MPLS forwarding (ultimately).
With an IP routing network, each /link/ /must/ have an
administratively configured IP address. (Hacks like Address-borrowing
(aka 'unnumbered') excepted).
I.e. The administrative overheads of IP only affect MPLS to the
extent it's used for the control-plane of MPLS, and those overheads
are fewer than if MPLS also needed IP for forwarding. MPLS is not
dependent on IP per se.
>> - lack of funky ways to route IP
>> - this one isn't really valid, can be implemented for IP..
>> Huh? I'm not even sure what this means.
Routing by ways /other/ than traditional lowest-metric path. E.g.
using additional metrics, like bandwidth, latency, reliability, etc..
as constraints on a traditional routing calculation, or using such
metrics in alternative calculations.
MPLS allows quite arbitrary routing, by being circuit-switched
(arbitrary association of packets->label and hence circuit, at
ingress). With circuit-switching you don't have to worry about nasty
routing loops, as you must with per-packet/per-hop routing - once the
circuit is setup, its there. (Doesn't mean circuit-switching is per
se better ;) ).
Paul Jakma paul at clubi.iepaul at jakma.org Key ID: 64A2FF6A
The universe is ruled by letting things take their course. It cannot be
ruled by interfering.
-- Chinese proverb
Maintained by the ILUG website team. The aim of Linux.ie is to
support and help commercial and private users of Linux in Ireland. You can
display ILUG news in your own webpages, read backend
information to find out how. Networking services kindly provided by HEAnet, server kindly donated by
Dell. Linux is a trademark of Linus Torvalds,
used with permission. No penguins were harmed in the production or maintenance
of this highly praised website. Looking for the
Indian Linux Users' Group? Try here. If you've read all this and aren't a lawyer: you should be!