We’ve taken the ATM and Frame Relay networks of yesteryear and applied the best parts of both to MPLS networking. In this video, you’ll learn about ATM and Frame Relay and how those networks helped build the MPLS networks of today.
<< Previous: SONET, DWDM, and CWDMNext: Satellite, ISDN, Cable, DSL, and Dialup >>
ATM stands for asynchronous transfer mode. It is a protocol that used Sonet as it’s transport mechanism, and it was designed use these 53-byte cells, 48 bytes would be your data, and then there would be a 5-byte routing header just before the data was sent over the ATM network. This was designed to constantly send these cells, which means that we had a very high speed, high throughput network, that allowed us to use real time protocols, and communicate using voice, or video, or data, over that same Sonet mechanism, using this ATM protocol.
We got maximum speeds of OC-192, and at the time, these were very, very high speed networks. We had to limit the speeds because every time you would pull the data off of the ATM network it had to be segmented to put on, and reassemble when it came off. So there was a bit of overhead associated with that. ATM did not really catch on, as we thought it might. We ended up having higher speed ethernet take over on the LAN, and ATM really was never able to make the jump on the WAN side.
On our wide area networks, we used to put up circuits from one point to the other in these point-to-point T1 or T3 links were nailed up and we’d have them up and running all the time, whether we were sending traffic over that link or not. We started to replace these point-to-point T1s with something that was a little more flexible called Frame Relay. In Frame Relay, we encapsulated our LAN traffic into these frame relay frames, and we sent them into what we called a frame relay cloud. We called it a cloud because our local router would put traffic into the cloud.
And this was all the providers world. We didn’t exactly know how the data was getting from one point to the other. All we knew was we would put traffic into the cloud, and then, magically, it would show up on the other side. And it was the responsibility of the provider in the middle to be able to make sure the data was traversing from one end to the other. Now we didn’t exactly understand what happened inside of the cloud, but we knew that we were using the bandwidth in the best possible way. Because if we weren’t using the connection, somebody else was using that connection, and that kept the cost down. Certainly less than paying for a T1 or T3 connection that was always being connected, all the time, whether we were putting data over it or not. The speeds generally range from 64 kilobits per second, through these T3, DS3 type speeds. So we were getting pretty good throughput, for the time, over these wide area networks using frame relay. Ultimately, we were getting better speeds and more flexibility through MPLS. So these days most of our frame relay networks have evolved and migrated into MPLS networks.
With MPLS, we took the best part of ATM, and the best part of Frame Relay, and we put all those things together to come up with a brand new wide area network mechanism. This allowed us to package up data, send it across the network, and we were able to direct where the traffic goes, based on something called an MPLS label. This also made it very easy for the end devices on the MPLS network to know exactly where the traffic was going because it had a very easy to follow label. This allowed us to put really anything inside of this transport mechanism. We could put IP information, we could put a ATM cells, we could put ethernet frames, and we were able to run this across a very common MPLS network.
So that really isn’t layer two, it’s not layer three, it’s somewhere in the middle if we’re looking to figure out where MPLS goes on in OSI model. You’re going to find MPLS in many places this is really the default for wide area networks these days. If you’re not running just a standard metro ethernet, it’s very common to see an MPLS network. This routing process on an MPLS network is called pushing and popping. When we’re putting information onto an MPLS network, we push a label onto it, and we put it into the MPLS cloud.
This is a good example of taking information coming from a customer edge router, connecting to the first provider edge router on the MPLS network, and pushing the route onto that connection. It then finds its way through the MPLS network, and on the outside, it pops off the labels and provides the regional traffic back to the customer edge router, and the process is repeated to go to the other side. Although we don’t see much native ATM or Frame Relay on our network any longer, we can really find the best parts of both of those contained in our MPLS wide area networks.