Whats next? As vendors put the 10 Gigabit Ethernet standards to bed, their focus is shifting to the next technological challenge. More bandwidth has always been needed, prompting suppliers to keep increasing Ethernets throughput by a factor of 10.
Unlike previous initiatives, this time, questions have arisen about the next performance multiple. “While there has been a lot of discussion about how to boost Ethernets transmission rate, I dont think a clear direction will emerge until the end of 2002,” said Chandra Kopparapu, director of product marketing at Foundry Networks Inc., in San Jose, Calif.
Thats because there are three options, each with its strengths and weaknesses. In the past few years, network equipment manufacturers have worked on a higher-speed version of SONET (Synchronous Optical Network), boosting its capacity from 10G bps to 40G bps, and that is having an impact on Ethernets future. One group of vendors wants to piggyback on that work and develop a 40G-bps version of Ethernet based largely on the OC-768 specification.
A second group wants to maintain the historical upgrade path, which has been done by multiples of 10. “[40G-bps] Ethernet does not offer enough of a bandwidth increase to interest all vendors,” said David Passmore, an analyst at the Sterling, Va., office of The Burton Group Corp., a networking consulting company.
A third option is to wait for development of the SONET OC-3072 specification, which promises 160G bps of bandwidth. However, work on that standard is just beginning, and product delivery would be three to five years away.
At present, there is no consensus on which of the three will emerge as the next step, and there is a distinct possibility that more than one option will be pursued. “A week ago, I thought I had a pretty good idea of how the current debate would play out, but today, Im not so sure,” said Mark Fishburn, vice president of technical strategy at Spirent Ltd., in London, a network test equipment supplier.
Another issue is changing Ethernets frame format, which packages data in 1,500-byte frames. One reason computers need high-speed microprocessors to support top Ethernet transmission rates is they have to break megabytes of data into small frames. As Ethernet transmission speeds rise, that task becomes harder and creates more processing overhead. When vendors were developing the Gigabit Ethernet standard and then with the 10 Gigabit Ethernet work, some suggested a potential solution: putting data in larger frames, say 8,000-byte frames, a technique called jumbo frames.
However, a group of engineers does not want to tinker with the frame size and believes it should remain the same. And while jumbo frames work well with large file transfers, they can be inefficient with voice communications because these transmissions come in small bursts, so individual frames may be underutilized.
As a result, jumbo frames were not incorporated into the previous standards but were offered as a proprietary add-on feature. “About 20 percent of our customers now use jumbo frames with Gigabit Ethernet transmissions,” said Foundrys Kopparapu.
Regardless of their position on jumbo frames, vendors seem to agree that work on a higher-speed version of Ethernet is not a pressing priority. “Currently, vendors are focusing on other opportunities, such as putting Gigabit Ethernet into the local telephone company loop that seems to offer them a better return on their investment than starting to develop a faster version of Ethernet,” said The Burton Groups Passmore.