* Metro Ethernet heading to be the de facto access interface
It seems that broader use of metro Ethernet services may be looming on the horizon as the next step for many enterprises. Most importantly, these services can be used not only for connecting campuses within a metro area, but also for accessing WAN services, including the Internet if your WAN connectivity includes an Internet-based VPN.
For years, access services to the WAN have been a major bottleneck in getting to the semi-infinite bandwidth available on the network. This has been a problem both in terms of available bandwidth and cost. In fact, for as long as we can remember, costs for getting to the WAN have been – in most cases – equal to or greater than costs for actual WAN transport.
Because of the technical and financial limitations, we have historically had to be quite concerned about overhead. For instance, one of the advantages of frame relay has always been the extremely low overhead, making this “skinny” protocol having only 5-bytes of overhead per PDU (protocol data unit, a.k.a., a generic “packet”) quite attractive.
This has been a historical problem with Ethernet services. Ethernet is a “fat” protocol with lots of overhead per PDU. However, overhead only matters when bandwidth is limited and/or expensive, so this is a quickly diminishing issue as the price per bit-per-second for metro Ethernet services continues to plummet.
Of course, there’s also a great benefit to metro Ethernet: simplicity. It’s a pretty safe bet that almost any data that’s flying around the LAN today is already in an Ethernet packet. (We’d love to hear from anybody who is still using Token Ring and/or ATM LAN.) Using metro Ethernet avoids the conversion from Ethernet to some other protocol for WAN transport, then reconverting back to Ethernet.
Let us know what you’re thinking about the future of metro Ethernet for your organization, and we’ll be glad to share the results. We’d especially like to hear what you think are primary drivers and primary inhibitors for implementation. We think this will become the de facto access interface. Do you?
