Trunking: A Middle Solution

The Ethernet market is now being driven by two concurrent trends: Gigabit Ethernet and Fast Ethernet. In the Gigabit Ethernet segment, shipments continue to rise as the technology moves out of its early adopter phase. But many buyers still prefer to remain on the sidelines, waiting for prices to drop. While they do, Gigabit Ethernet is a viable solution only for the most needy and price- insensitive. Its mass-market phase is still in the future.

The Fast Ethernet market is just the opposite. It is now entering a commodity phase, driven, in part, by many new suppliers offering very good products at rock-bottom prices. These vendors have built Fast Ethernet L2 switches around merchant-chip technology, a capability that gives any company with a good engineering staff and established channels a shot at some market share in this space.

In addition, the top-tier networking providers have not fallen asleep. Since their first product introductions, they have announced wave after wave of enhancements that improve Fast Ethernet performance while lowering prices. In combination, this aggressive product activity by first- and second-tier suppliers has driven the Fast Ethernet market into its commodity phase, making this technology cost-effective for mass deployment in enterprise workgroups and small and medium-sized businesses.

So what's a buyer to do when both price and performance are needed? On the one hand, Gigabit Ethernet transmission capacity is available, but only at a premium price. On the other hand, users can buy Fast Ethernet technology at rock-bottom prices, but only if they are willing to live with its 100-Mbps speed. Ethernet buyers needing both have been torn by this price/performance dilemma. Recently, however, @@Itrunking@@SR has come into play as a solution to this problem.

Trunking is the ability to gang together multiple Fast Ethernet ports to enable them to work together like one logical channel. Users can bring together four full-duplex ports, for example, into one 800-Mbps link, effectively coming very close to Gigabit Ethernet's speed but at the price of four Fast Ethernet ports.

Trunking is not a bare-bones technology. It supports load balancing that effectively distributes traffic across all the links, thereby providing parallel paths for all data. In the event of a link failure, the traffic is automatically redirected to the remaining links -- all done in millisecond convergence time without the need for user intervention.

Over the years, trunking has been used in many different sectors of the communications industry. In the LAN market, for example, Kalpana, bought in 1994 by Cisco Systems Inc., was the first to use it in its products way back in the early days of switching. Trunking, in fact, is not even new to the Fast Ethernet market. It has been available for some time but has been positioned primarily as a switch-to-switch connection technology. This restrictive positioning, of course, has lowered its appeal as an alternative to Gigabit Ethernet. Until it could be used for both switch-to-switch @@Iand@@SR switch-to-server connections, Fast Ethernet with trunking was viewed as a half-hearted alternative to Gigabit Ethernet -- regardless of its price advantages.

This view is starting to change as server manufacturers step forward with host-side adapters capable of working with the trunking on Fast Ethernet switches. Because of these adapters, Fast Ethernet with trunking is finally coming into its own as a low-cost alternative to Gigabit Ethernet. With a speed range between 200 and 800 Mbps, a Fast Ethernet switch with trunking can be used in either switch-to-switch or switch-to-server environments. Additionally, this solution has attractive investment protection. While performance can be bumped to near-gigabit speed, the solution does not call for the junking of any Fast Ethernet equipment.

There's more good news on the testing front. Without a doubt, buyers should be concerned about the interoperability between switches and servers that are manufactured by different suppliers. Tests by independent laboratories, however, show that there is little need for concern on this point. To date, lab tests have proved switch/server interoperability when trunking is used between switches and servers. And there is more good news on the performance front. Although trunking may claim an 800 Mbps capacity, the real issue is the bandwidth utilization. That is, how much of that capacity is actually used? Once again, independent lab tests show that bandwidth utilization is very high when trunking is used between switches and servers.

In short, switch-to-server connectivity puts in place the final piece that makes Fast Ethernet with trunking a viable alternative to Gigabit Ethernet. Buyers who had faced the Gigabit Ethernet/high price vs. Fast Ethernet/low performance dilemma now have a middle solution that offers the best elements of both.

Sam Alunni is vice president of networking at Sterling Research (Sterling, Mass.). Contact him at alunni@sterlingresearch.com.