JOHN WILLIAMSON 400G SPEEDS
Similar to 100G, there will be a range of package styles available for 400G, as technology advances package sizes and power consumption will reduce.
capacity between successive generations of technology. For the future, there is an expectation that Ethernet will start to scale on a more granular, flexible and intelligently managed basis. In this context one relevant initiative is the work on Flexible Ethernet (FlexE). announced the completion of the FlexE Implementation Agreement (IA) for DCI. This accord is designed to provide a generic mechanism for supporting a variety of Ethernet MAC rates that may or may not correspond to any existing Ethernet PHY rate. This includes MAC rates that are both greater than and less than the Ethernet PHY rates used to carry FlexE. “The FlexE IA allows Ethernet to be used more efficiently in response to ever changing technologies, link rates, and bandwidth needs,” according to Scott Irwin of MoSys and the OIF’s Physical and Link Layer Working Group – Protocol Vice Chair. Even so, more big beast optical capacity could be just over the horizon with Terabit systems becoming the topic of industry discussion. “It seems a long way off, but some of the things we’re doing are examining how you might get there with the technology we have,” states Carter. Optical muxing or operation in parallel environments are among the possibilities. In March 2016 the Optical Internetworking Forum (OIF)
40 billionth of a second, and we’ve got many of the signals working in parallel,” observes Brooks. “400G is going to be a very big inflection point for the whole ecosystem.” SLOW BUT SURE Notwithstanding its game-changing potential, there is some consensus that 400G will not spell the end of the line for lower speed optical communication systems. Fischer observes that there will be continued demand for, and operation of, 1G, 10G and even below Ethernet solutions for particular applications and services that are then aggregated before being actually be put onto a metro or long-haul transmission system. Brooks can envisage a 400G ecosystem in which you have servers, perhaps working at 25G, feeding top of racks working at 100G which, in turn, feed into centre racks running at 400G. “All of these rates have a role to play,” he argues. One exception here, at least in terms of new CapEx, may be 40G systems (see panel “Forty neiners”). “I don’t expect much investment in 40G – investment in lower rates will be in 25G, 100G and even some 50G,” says Brooks. NEXT UP In the recent past optical communications development has been characterised, inter alia , by fairly big step increases in speed/
variables
such as number of
modulation states, baud rates, spectral efficiencies and
transmission distances. Most strikingly, 400G looks to be a major stretch from an operational perspective. The scope of the technology envisages the replacement of binary modulation with four-level Pulse Amplitude Modulation (PAM4). Moving up from the 4 x 25 Gbits/s of 100G, 400G has 8 parallel channels of slightly over 25 Gbd. “It’s phenomenally fast – a signal is turning on and off in one
www.opticalconnectionsnews.com
19
ISSUE 7 | Q3 2016
Made with FlippingBook - Online catalogs