Following this tradition the subsequent Ethernet standard with 100Gb data rate should be released by end of 2007. This is not likely to be the case. In the IEEE community the Higher Speed Study Group ( HSSG ) is presently discussing several technical scenarios for 100Gb Ethernet as prework for an official standardization task force. Among the discussed solutions are 10x10Gb, 5x20Gb and 4x25Gb platforms with several optical reaches, each of them using optical inverse-multiplexing.
This delay with respect to the Ethernet tradition may be explained by the fact that the leap from 10Gb to 100Gb implies several technological challenges.
In order to save costs on the optical side a four channel solution would in general be preferred to a 10 channel solution. But this implies the development of new lasers providing good temperature stability at higher data rates (3dB bandwidth of 20-25Gb) combined with a relatively high optical output power. It is unclear to which data rates directly modulated lasers may be extended or whether external modulated lasers are the better choice. On the other hand, a 10 channel solution, which would benefit from the established laser technology, suffers from higher costs, stronger power consumption and larger package size.
On the electrical side the use of the well established and inexpensive FR4 PCB material limits the data rate of electrical signal processing. To overcome this limitation the data rate may be either de-multiplexed electrically or new technologies and material systems for electronic chips, PCB material and connectors may be applied.
Within the Optical Internetworking Forum ( OIF ) the Physical Link Layer ( PLL ) Working Group investigates and discusses a new electrical interface called CEI-25 which introduces technological improvements on the electrical printed board material ( improved FR4 ) and design with the objective to provide 100Gb (25Gb per channel) on the backplane. MergeOptics is a member of OIF. |