10Gbps networking is coming soon

10G is coming soon - in the backbone as well as the horizontal. The SYSTIMAX SCS End-User Survey conducted in October 2002 revealed that 59% of end-users expect to run 10G in the backbone within the next five years and 26% of end-users expect to run 10G in the horizontal within five years. In an effort to meet this expectation, IEEE continues to launch efforts to broaden the choices for more cost-effective solutions.

10G on twisted pair gets the green light

At the November 2003 IEEE meeting, the green light was given to start an official project to develop 10GBASE-T, a 10Gbps Ethernet on twisted pair standard that is the first to require Category 6 cabling as a minimum. The target completion date of the standard is 2006. This development highlights the value of installing better UTP to facilitate the lowest total solution cost for 10Gbps networking in the horizontal.

10G on legacy multimode fibre stumbling forward

Limited availability of 10GBASE-LX4 transceivers sparks novel development to compensate for mode delay in legacy fibre cabling via electronics. A study group has been formed to review the options for Electronic Dispersion Compensation to overcome the deficiencies of legacy fibre cabling. In the meantime, the lowest-cost 10GBASE-SR modules are beginning to ship in volume, highlighting the value offered by laser optimised (OM3) multimode fibre for new installations.

A big step forward for twisted pair cabling: IEEE 802.3 committee unanimously supports formation of the 10GBASE-T Task Force

In November 2002, IEEE 802.3 approved the formation of a study group to investigate the feasibility of standardising 10Gbps transmission over twisted pair cabling. At the November 2003 requirements (broad market potential, compatibility, distinct identity, technical feasibility, economic feasibility) were satisfied, IEEE 802.3 unanimously approved the project authorisation request forming the 10GBASE-T "802.3an" Task Force.

Due to the expected complexity of electronics to support 10GBASE-T, an early objective to support Category 5e was dropped, and the exact maximum distance over minimally compliant Category 6 cabling is still uncertain. The final mandatory objective for the project is "at least 55m to 100m" over Category 6 or better cabling. Many expect that, as the clever chip designers further dive into this project, novel techniques will be developed which will increase the minimum guaranteed distance over generic Category 6 cabling. Margin over the minimum Category 6 specifications will also extend the distance capability.

A SYSTIMAX Labs presentation at the March meeting was the first to show the capability to dramatically improve UTP performance for 10G. It became clear in November that other vendors, conscious of the worldwide preference for unshielded cabling, are now also showing progress in UTP cable design and are supportive of a UTP initiative. This is great news for the 10GBASE-T effort. Most are now confident that a combination of improvements in chip design and cabling with guaranteed margin above Category 6 will make 10Gbps transmission over cost-effective UTP cabling a reality. This was essential to achieve "Broad Market Potential" since a solution requiring Category 7 cabling was not viewed as viable due to very low market penetration, component and installation cost, bulk and grounding concerns. Although it will be too early to guarantee distance support until the final coding scheme is chosen, SYSTIMAX Labs fully expects that customers with certified GigaSPEED XL installations will enjoy longer distance support and/or better reliability than the final distances supported by the IEEE 10GBASE-T for Category 6 cabling.

10G legacy multimode fibre solution still stumbling and approaching re-design

A new study group will explore a more cost-effective 10G solution for legacy multimode

Four distinct 10G fibre solutions have been standardised to address different 10G market segments. The solutions vary in distance support as well as fibre type support. 10GBASE-LR and 10GBASE-ER are designed to run on singlemode fibre for long distances (eg campus and metro networks).

Although transceivers to support singlemode (10GBASE-LR and 10GBASE-ER) and laser optimised multimode fibre (10GBASE-SR) are already shipping in volume, the original intent to provide cost-effective support for the installed base of legacy 62.5 and 50 micron fibre is stalling due to scarce availability of the complex 10GBASE-LX4 transceivers.

10GBASE-LX4 splits the signal into four 2.5Gbps data streams, utilises Coarse Wave Division Multiplexing (CWDM) to transmit each data stream at a different wavelength, and recombines the data streams at the receiver. In addition to the multiwavelength complexity, 10GBASE-LX4 requires costly optical multiplexing/de-multiplexing components. As a result, transceiver vendors have found it extremely challenging to deliver 10GBASE-LX4 transceivers cost-effectively. In an attempt to overcome this issue and improve the potential choices to support the installed base, a new study group has been formed to explore a novel technique to utilise serial 1310nm lasers with electronic dispersion compensation (EDC). EDC is designed to compensate for the severe pulse splitting in legacy multimode fibres caused by poor Differential Mode Delay (DMD). The use of sophisticated compensation techniques to correct pulse splitting should not add too much cost to the existing 10GBASE-LR option intended for singlemode. So is it back to the drawing board for legacy multimode support? Time will tell if either LX4 or the new EDC option (if developed into a standard) are a better option than replacing legacy multimode.

In the meantime, large equipment vendors are already shipping 10GBASE-SR ports for the lowest cost system option over laser optimised (OM3) fibre for channels up to 300 meters and 10GBASE-LR for longer channels (where singlemode fibre becomes the only option, and the transceivers command a high premium). Having anticipated the difficulties of delivering 10G over legacy multimode fibre, SYSTIMAX Labs developed the LazrSPEED Solution in 1999, and scores of registered SYSTIMAX installations worldwide are ready to take advantage of the lowest-cost 10GBASE-SR option, having chosen LazrSPEED 150, LazrSPEED 300 or LazrSPEED 550 depending on distance requirements.