MPLS router

[Qacer]

Regarding MPLS routers, I keep hearing about them being part of the core network. By core, does this also translate to a WAN environment (e.g. each router is separated by hundreds or thousands of miles)?

Also, what transport technology is used to connect each router to each other? Are MPLS routers normally connected with long fiber lines using an Internet Protocol over Wave Division Multiplexing (IP over WDM) technology? Or do they actually "ride" on a SONET/SDH link?

Thanks!

 

[imagoon]

MPLS is not that differnet from a vpn point-to-point run over either the Internet or inside another corps WAN. IE AT&T MPLS will give you a router at one end, pack the packets with routing data and send it through their own back bones to the other end where that router strips the routing data and then handles it. The main difference is that ATT can over subscribe the lines a bit which makes it cheaper than say a T1 point-to-point. The T1 will always have 1.536mbps even if you don't send anything. MPLS links (inside ATT) will route data even if your not using your share. ATT will tend to have small pings because they don't go crazy and sell it at 1mb:15 users resulting in say 5ms pings. Something like Megapath has a slightly cheaper price but will only net you 10ms pings.

EDIT

Think of it this way, you have a huge switch that you can configure vlans on, port 1 and 27 can be on vlan 2 while 2 and 28 are vlan 3. They can't talk to the other vlans because the switch blocks it. However is the other port is on another switch, all the vlans can be trunked up and over to the other switch where it could put port 3 on vlan 2 and port 29 on vlan3. That trunk is shared. There is some back end routing in there that actually makes all the wan connections "connect."

MPLS is a packet switched protocol that rides between layer 2 and 3. It also can encapsulate ATM / SONET and Ethernet frames.

From my understanding, it is basically the "big brother" of ATM with some extra things thrown in to make it easier to handle.

More info if you can excuse the wiki link:

http://en.wikipedia.org/wiki/M...otocol_Label_Switching

 

[Cooky]

MPLS is actually a tag switching technology (NOT packet switched) that can be implemented over frame or cell-based layer 2 infrastructures.

The packets or cells are switched based on labels, which are distributed over TDP or LDP.
The source & destination routers form a LSP (label switched path), and the routers in between don't need to run BGP at all...they just switch traffic according to the labels.

Edit:
The physical infrastructure could be anything...frame-relay, ATM, SONET, ethernet...

When people say MPLS is part of the core, they probably meant it's part of a provider's core.
Very few people run MPLS on the campus core, since the same purpose can be achieved w/ VRF-lite.

The way they differentiate different customers is by the use of RD (route distinguisher).
So customer A's 192.168.1.0/24 is different from customer B's 192.168.1.0/24, and yet they're both routed on the same provider edge router.

 

[Qacer]

Thanks!

Wikipedia isn't the truth, but statements such as this one "With the advent of MPLS, VPN and dedicated broadband services such as cable modem and DSL, the end may loom for the frame relay protocol and encapsulation" leads to confusion. MPLS is considered to function at Layer 2.5 using label switching and stacking, so it's not necessarily tied up to the physical infrastructure. I'm trying to find out what physical infrastructure MPLS is normally implemented on. If the Wikipedia statement is true, then it implies MPLS uses its own transport technology.

 

[Emulex]

wow people still use the old frame relay with CIR?? in the old school days folks could use a switched service to "dial-up" i suppose other destinations.

i remember weighing the cost of a T-1 from mci versus a frame relay 1.5 megabit with 768kbps CIR and deciding the FRAD was not the most interesting device on the planet.

old times.

now you can have comcast business with many ip's (50 meg down,10 up) as a solid static backup line for not much.

 

[Pantlegz]

Originally posted by: Emulex
wow people still use the old frame relay with CIR?? in the old school days folks could use a switched service to "dial-up" i suppose other destinations.

i remember weighing the cost of a T-1 from mci versus a frame relay 1.5 megabit with 768kbps CIR and deciding the FRAD was not the most interesting device on the planet.

old times.

now you can have comcast business with many ip's (50 meg down,10 up) as a solid static backup line for not much.

Thanks for this, I was just learning about frame relay and though it sounded absurd and more expensive,in most cases, than a business account with an isp. My instructor couldn't come up with a situation that made sense to me to use frame relay.

 

[drebo]

There are lots of good reasons to use frame relay, or indeed any point to point or leased line technology, over "a business account with an isp." When downtime costs thousands, if not millions, of dollars an hour, it's worth it to spend the few hundred dollars per month extra on the more reliable connection.

Businesses who rely on the internet or on their connections to the central office should not be using consumer broadband connections (and despite what Comcast/Charter/AT&T DSL may say, all of their "business" services are identical to their consumer services in both form and function) to achieve those connections. MPLS does away with the need for frame relay, sure, but it's not going to be any cheaper, and it honestly doesn't need to be.

 

[ScottMac]

Originally posted by: drebo
There are lots of good reasons to use frame relay, or indeed any point to point or leased line technology, over "a business account with an isp." When downtime costs thousands, if not millions, of dollars an hour, it's worth it to spend the few hundred dollars per month extra on the more reliable connection.

Businesses who rely on the internet or on their connections to the central office should not be using consumer broadband connections (and despite what Comcast/Charter/AT&T DSL may say, all of their "business" services are identical to their consumer services in both form and function) to achieve those connections. MPLS does away with the need for frame relay, sure, but it's not going to be any cheaper, and it honestly doesn't need to be.

MPLS would operate on top of frame relay (or any other layer two). MPLS operates at layer three and above, frame-relay is a layer two technology.

Frame also permits rapid switchover on network faults or for emergency cut-over to a backup data center. It is also easy to establish a single (or redundant) aggregated link from a constellation of field locations.

I haven't heard if the CableCos are offering MPLS on their networks or not. Since many/most use other carrier's networks for intermediate and long-haul, I'm not sure a label-switched path is available ... maybe they're tunneling ... I don't know.

 

[Emulex]

every point is a point of failure leased, shared, renting, owned.

you must remove every single point of failure these days. null routes or bad routes/peers, ddos, business failure are not uncommon if you've been around you've seen it all.

why companies host their servers at NAP's in multiple datacenters nowadays.