MPLS/IP switching

[solariusx3]

As above. why is it gaining popularity these days among companies and isp's? wiked it but still don't quite understand.

 

[Cooky]

MPLS gives you more control over your traffic than the traditional frame-relay circuits.
One good example is QoS.

 

[jlazzaro]

MPLS works by tagging packets with an identifier (a label) to distinguish the LSPs. When a packet is received, the router uses this label (and sometimes also the link over which it was received) to identify the LSP. It then looks up the LSP in its own forwarding table to determine the best link over which to forward the packet, and the label to use on this next hop.

A different label is used for each hop, and it is chosen by the router or switch performing the forwarding operation. This allows the use of very fast and simple forwarding engines, as the router can select the label to minimize processing.

The big advantage to MPLS is a fully meshed topology. In a typical hub-spoke scenario, youd have your core (hub) site with remote sites attached to it. If one remote site wanted to communicate with another, it would have to go through the hub. With MPLS, you have direct Any-to-Any connectivity. Even if the hub was down, you would have connectivity between your all your other sites because the ISP routes your traffic based on labels directly to your destination. This is even more useful when you take into consideration other technologies like VoIP and Video Conferencing.

High speed, low latency, functional, flexible, expandable Comes at a cost though...our MPLS circuits on average cost about double our tradition point-to-point T1's.

 

[spidey07]

jlazzaro,

weird that mpls cost more. It should cost less.

 

[cmetz]

solariusx3, MPLS is used for different reasons by ISPs and enterprises.

ISPs use MPLS as a traffic engineering tool, primarily as a way to work around the limitations of BGP and existing Cisco/Juniper routers. In particular, BGP has this tendency in practice to like to dump a lot of routes onto one link and leave another mostly idle, so MPLS allows ISPs to create extra, fictitious links and then massage BGP metrics to cause "better" balancing between physical links to happen. Also, those same fictitious links allow ISPs to gather utilization metrics between "sites" that they can feed into their capacity planning tools and determine when more physical capacity or a BGP metric rebalancing would be helpful.

Enterprises use MPLS as a way to get cheap virtual MAN/WAN circuits. If I have to connect two sites on different coasts with a 10Mb/s circuit, chances are that MPLS is going to be the cheapest way for me to do that. I can also use MPLS to connect three sites in three very different locations and only have one tail circuit per site, and let the network set up VC paths from each site to the other two (so a full mesh).

In both cases, MPLS is basically a direct replacement for ATM, which in turn is basically a direct replacement for Frame Relay. The technologies themselves have important differences but few people use them in ways where the technology itself would matter much, they're just looking for something that solves their problem. Frame Relay was a good technology IMO and it was mostly ATM's better PR and the exit of Cascade that killed it. ATM was a horrible technology and the combination of freaky telco circuit-switched religion and academic queuing-theory QoS guys made it a huge mess - most notably it simply slammed into a wall in terms of performance scaling on the SAR. MPLS is in many ways simpler and more sane, and happens to be developed by mostly Cisco folks (tag switching) who are focused on the problem of carrying IP between sites and not in trying to do native MPLS to the desktop or some other silliness. But a lot of the same problem people and problem features from ATM crept into MPLS, too, so if you feel the need to make your system more complex than it needs to be, you can do that

 

[cmetz]

spidey07, within a LATA, MPLS probably would cost more than a point-to-point T1. Once you start going a long distance, that would reverse.

 

[jlazzaro]

Originally posted by: cmetz
spidey07, within a LATA, MPLS probably would cost more than a point-to-point T1. Once you start going a long distance, that would reverse.

indeed sir. we saved big on some crazy T1 mileage fees, but that flat $415 subscription MRC for every circuit is what kills it.

 

[spidey07]

Originally posted by: cmetz
spidey07, within a LATA, MPLS probably would cost more than a point-to-point T1. Once you start going a long distance, that would reverse.

true. If you stay withing your LATA you can get TDM circuits for dirt cheap.

 

[solariusx3]

ah thanks guys its starting get together now...reps for those who contributed. btw wats LATA ?

 

[jlazzaro]

Local access and transport area...an area that is the responsibility of a LEC (local exchange carrier).

 

[cmetz]

There are tarriff / regulatory reasons why crossing a LATA gets expensive. Also, some circuits won't be available from some carriers inter-LATA. It's a silly historical thing that bites you all the time.

You can have two circuits of the same type and same distance, one intra-LATA and one inter-LATA, and they can have wildly different costs. The intra-LATA circuits are usually cheaper, but even that isn't always true. At a rough conceptual level, you can think of intra-LATA as "local" and inter-LATA as "long distance." That was more or less the original intent.

Many US metro areas have grown such that what was once rural is now suburb, so it's now surprisingly common to have multiple LATAs in a major metro area. It's therefore handy to know where those boundary lines are drawn. You might not think of a circuit to the next town over as being "long distance" but it might be inter-LATA.