MPLS and Ethernet: Seven Things You Need To Know
Ethernet has come a long way in its eventful 30-year life. As the de facto LAN standard, Ethernet is now being deployed in the WAN. In parallel, the demand for MPLS-based technology is pressing service providers to deploy Ethernet as an access or "handoff" technology. Enterprises stand to gain from this simple LAN-to-WAN interface by freeing themselves to deploy increasingly advanced services while outsourcing the smart management stuff to service providers.
As with any technology, it's useful for enterprises as well as service providers to understand VPLSthe key technology that facilitates Ethernet-based WANs. As we'll see, not only does VPLS leverage multiprotocol label switching (MPLS); MPLS also leverages tested-and-tried IP technology. This is a solid foundation.
Ethernet = Simplicity
As enterprises continue down the road of outsourcing and focusing on areas of core expertise, there's a growing need for network simplicity. Ethernet has provided this simplicity in the enterprise LAN for many years now. The limitations of the collision-based Ethernet model have been reduced by the use of Ethernet switches. Even service providers have deployed switched Ethernet, although it appears that this technology is running out of steam for the new generation of real-time services. Despite this limitation, Ethernet is even appearing in optical transport networksthere appears to be no limit to the applicability of Ethernet. You might say we're moving toward Planet Ethernet!
Network services are being delivered across the WAN to branch offices, remote workers, and even customers. Using Ethernet end-to-end for all such services helps overworked network managers get out a little more. In Figure 1, Link A, Link B, and Link C are all Ethernet-based and serve to connect the three different site types so that they appear to be part of one big LAN, even though they're interconnected by a service provider network.
Figure 1 An IP/MPLS virtual private network.
Each customer site in Figure 1 hosts a customer edge (CE) devicenormally a router. The CE is linked to an adjacent provider edge (PE) devicein many cases, a multiservice router. The merit of this arrangement is that the CE nodes can be simple IP routers, whereas the PEs are complex and expensive multitechnology devices.
The services provided to the users in the customer sites give rise to traffic across Links A, B, and C. Given that these services are increasingly a mixture of real-time (such as voice over IP) and non-real-time (data), it's essential that the underlying network technology (that is, Ethernet) supports this capability. How is this done? The answer lies in the way Ethernet can now deliver different levels of quality of service (QoS).