From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Thu Jun 12 2003 - 15:52:06 GMT-3
At 10:35 AM -0400 6/12/03, Peter van Oene wrote:
>At 06:48 AM 6/12/2003 -0500, Jason Cash wrote:
>>Well, the idea being, that at some point everybody was a dummy before they
>>became an expert. I have absolutely no undetstanding of MPLS, hence, I
>>asked for links to learn about it. I can tell you this, even Jimi Hendrix
>>sucked the first time he picked up a guitar.
>
>I'd go to the mpls working group at www.ietf.org and read some
>architecture rfcs to start. rfc 3031 would be an excellent start.
Actually, before that, it would help to get something of a big
picture. This happened to come up in some recent correspondence with
a very expert colleague who was having trouble grasping MPLS:
At 12:47 PM -0400 6/10/03, Howard C. Berkowitz wrote:
>You have an advantage, then, of not having preconceptions. Cisco
>Learning presentations on MPLS usually are _incredibly_ bad. They
>almost always spend their time about forwarding and ignore how the
>MPLS paths are set up.
>
>I would recommend backing off a little, and being sure you
>internalize the concept of "sub-IP" protocols of which (G)MPLS is
>one. Also, internalize the concepts of the ISO Internal
>Organization of the Network Layer.
>
>(G)MPLS, did I say? Yep. Generalized MPLS.
>
>"Ordinary" MPLS is a means of imposing a circuit-switched discipline
>onto a packet-switched network, but without the baggage of ATM.
>Circuit-switching, in turn, lets you reserve resources, both for QoS
>and for failover.
>
>MPLS and IP are far better coupled than ATM and IP. Ordinary MPLS is
>tunnel-based, so IP protocols can be used to set up the tunnels.
>Supplementary MPLS protocols then distribute the MPLS-specific layer
>management information.
>
>Generalized MPLS adds the capability of managing non-packet-switched
>services, like wavelengths/lambdas, TDM timeslots, and spatial
>relationships (e.g., ports on a physical switch).
>
>
>>I interpret dumb as not being well-versed in a subject, whereas an idiot
>>would be incapable of learning. So I would rather be dumb than an idiot any
>>day of the week.
I don't know whether to call it a "dummy" principle, but in my
earlier programming days, I tended get stuck in one (or a very small
number) of language features before I could "break through" and start
using it. In IBM 360 assembler, it was the idea that called
subroutine was passed a pointer to a list of parameters, not the
parameters themselves. In C, it was the simple way to establish the
beginning and end of a function.
In certification, there's a lemming-like pressure to get into
configuration before understanding why you want to do something --
not necessarily protocol theory, but protocol/feature applicability,
and how the protocol designers expected it to be used. This is going
on in the OSPF partitioning thread, to which I'm preparing another
response.
OTOH, the applicability and use of MPLS is definitely what I would
consider an advanced topic. To have a real understanding of it, you
need a deep internal understanding of some of the limitations of IP
and ATM and how MPLS overcomes them. Very soon if not now, you will
also need to know the limitations of wavelength "routing", TDM
including SONET, and DACS-like devices and how GMPLS solves them.
It's also essential to understand that (G)MPLS is sub-IP and both
complementary and tightly coupled to IP for its control.
I'm a little puzzled why Cisco is pushing it as part of general R&S,
because I consider it primarily a carrier technology. Yes, I suppose
you can argue that enterprise-oriented people should understand what
the carriers do with their data, but, if that's the argument, why
doesn't R&S deal with a realistic range of BGP multihoming scenarios?
>>
>>-----Original Message-----
>>From: Jason Graun [mailto:jgraun@attbi.com]
>>Sent: Wednesday, June 11, 2003 11:30 PM
>>To: 'Jason Cash'; ccielab@groupstudy.com
>>
>>MPLS and dummies should never be used on the same page, let alone the same
>>infrastructure. Why is there a notion in the IT field that anybody
>>including "dummies" can be an "expert" and make a lot of money? Just a
>>thought.
>>
>>-----Original Message-----
>>From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
>>Jason Cash
>>Sent: Wednesday, June 11, 2003 4:46 PM
>>To: ccielab@groupstudy.com
>>Subject: R&S MPLS
>>
>>Can someone provide me with a few links to MPLS case studies and/or examples
>>for cisco networks. A "MPLS for dummies" would be nice.
This archive was generated by hypermail 2.1.4 : Fri Jul 04 2003 - 11:10:57 GMT-3