Global Sources
EE Times-India
Stay in touch with EE Times India
EE Times-India > Networks

Ethernet ties LANs in wide areas

Posted: 01 May 2001     Print Version  Bookmark and Share

Keywords:ethernet  layer 2  layer 3  lan  wan 

The transparent LAN is a fairly mature technology, with numerous methods of transport delivery available. In the last two years, its 10Mbps and 100Mbps offerings have been fairly successful. Further technological advances, such as the extension of Ethernet standards to incorporate a 1Gbps version and the decreasing cost of 10Mbps and 100Mbps Ethernet circuits, enabled a seismic change to the WAN landscape.

One of the early uses of metro Ethernet services was to attack the matter of LAN performance across multiple sites within a metropolitan area. Traditionally, each LAN in a building or campus was a high-performance island connected to other high-performance islands with a much slower WAN connection. It is not unusual to find companies with 1Gbps LAN backbones connected by DS3 (45Mbps) WAN links, or even 100Mbps LAN backbones connected by DS1 (1.5Mbps) WAN links. Clearly, such an approach has a severe impact on performance as a whole. The concern becomes even more important as companies migrate to a "switched Ethernet-to-the-desktop" topology where each desktop has a dedicated 10Mbps or more to the switch.

Traditionally, each location would utilize Layer 3 WAN transport and return to Layer 2 on the terminating link. That arrangement reflected distance constraints that applied to native Ethernet and, until very recently, severely limited the range (fewer than 1,000m) that could be bridged without a Layer 3 solution. The initial solution would have been a private line between the two locations. But a private line is an expensive dedicated access service; in fact, the premium pricing of private lines resulted in the development of frame relay and atm partly to gain similar functionality at a reduced price.

Private line, frame relay and ATM denominate their services in such standard telecom increments as T1 (1.5Mbps), T3 (45Mbps) and OC-3 (155Mbps). But because none of those increments has a one-to-one correspondence with LAN backbone speeds (10Mbps and 100Mbps and 1Gbps), it is not easy to adapt them to provide seamless transport from LAN to LAN.

An alternative approach is to use Ethernet as a carrier-class, Layer 2 transport service that is scalable, dedicated, protected and provides guaranteed bandwidth. The customer-interface requirement is a standard IEEE 10Base-T, 100Base-TX, 1000Base-LX or 1000Base-SX connection. A service provider could install all the transport equipment at the customer premises and charge a flat monthly fee for the service.

The emergence of Ethernet-over-IP technology (a Layer 3 solution) simplifies the provisioning of IP transport and enables an alternative approach to the same problem. Such offerings have been marketed as Gigabit Ethernet and sold in denominations of 1Mbps.

Actually, these services are not so much a Gigabit Ethernet or Ethernet solution as they are IP-based virtual private networks with Ethernet connections. That makes them very useful for general Layer 3 applications but not as efficient or reliable as direct Layer-2-to-Layer-2 connectivity. Even if the providers are not oversubscribing their IP networks, overhead and latencies will be higher for Layer 3 than for Layer 2.

So far, the discussion has centered on a basic example connecting two geographically separate locations at LAN backbone speeds. A more complex question is what happens when there are more than two sites. An approach is required that serves the market from two perspectives: point-to-point and any-to-any connectivity options.

Suppose a customer has four sites on metropolitan fiber. It would like all of the locations to "see" all the other locations. In that case, the customer hands off the standard IEEE Ethernet connections at each site and the service provider manages the connectivity for the customer by providing dedicated, redundant Ethernet circuits. In that fashion, the benefits of switched Ethernet-such as full-duplex dedicated bandwidth without frame collisions-allow for very efficient connectivity between sites, much lower than a Layer 3 link would allow.

The topology of these networks can vary from customer to customer. In the star configuration, a customer's switch complex at one site provides connectivity to the balance of sites. The ring configuration, on the other hand, relies on the customer's having switches at each location to provide connectivity redundancy.

All of this discussion would be irrelevant if the price of the service was even similar to those of private line, frame relay and ATM transport. In fact, Ethernet services are much less expensive for the customer on a per-bit basis. In many cases, customers can upgrade their dedicated bandwidth to LAN backbone speeds and pay less than they would on their traditional transport.

Therefore, the economics of Ethernet will drive a substantial wedge into the existing installed base of private line, frame relay and ATM. Another factor that will contribute to the success of Ethernet services is the ease of scalability that exists when customers purchase additional bandwidth. Additional full-rate circuits-more than 60 1Gbps circuits per fiber pair can be provided in days rather than months. Comparable traditional circuits can take as long as six months to set up and have less bandwidth capacity as well.

Dan Kalin

XO Communications Inc.

Comment on "Ethernet ties LANs in wide areas"
*  You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.


Go to top             Connect on Facebook      Follow us on Twitter      Follow us on Orkut

Back to Top