Recently, I read a very interesting article on the IEEE drafting a new 10-Gb Ethernet standard. Looking ahead, this is more than just "interesting" to MRVC and NAC. As you know, for LANs and enterprises, the world is predominantly Ethernet. Having a level of Ethernet capable of WAN level performance could have a major impact on Metro Networks and I think the trend will play right into MRVC's and NAC's strengths.
It would be so easy and cost effective if WAN Metro networks could just deliver high speed Ethernet to customers, which of course is where NAC is going. I think MRVC and NAC will be leaders in this critical new area. Consider these scraps:
Noam said in the Q3 CC that they were near to both OC-48 (2.5 Gbps) and OC-192 (10Gbps) transeivers, I think he stated "within 4 months".
From the IEEE article:
"A first draft of the 10-Gbit spec is due next September, and proponents said they hope to bring a final draft to IEEE ballot in March 2001. The full specification is likely two or three years away, but panelists at the conference said they expect to see products based on the partial spec as early as next year."
From the NAC website:
"Speed limitation on native data interfaces such as Gigabit Ethernet will soon be removed by the advent of 2.5 or 10 Gbps interfaces. New Access will lead this innovation with its 2.5 Gbps native IP framing structure for the internal transport links. Thus the system allows for direct throughput of 2.5 Gbps per DWDM transceiver with user interfaces of either two Gigabit Ethernet lines or twenty 10/100 lines for each wavelength. Optimized for the demands of internet traffic, MetroFusion reproduces features available in legacy SONET networks using intelligent DWDM systems that can scale to much higher bandwidths."
From the MRVC 3012 press release last January:
"The GFS3012 also has a unique module that combines 4 Gigabit Ethernet data streams onto one fiber for transmission up to 50 kilometers. This Wave Division Multiplexing (WDM) feature provides interswitch links of up to 4 Gigabits in bandwidth. Using WDM between buildings, network managers can insure bandwidth availability for all users regardless of their location in the Metropolitan Area Network. "
Noam also said in the Q3 CC that Zuma will offer order of magnitude improvements over MRVC current Gigabit products.
Back to the IEEE article:
"The IEEE 802.3 working group has set an aggressive schedule for completion of 10-gigabit Ethernet, betting that the souped-up standard will be the first Ethernet version to establish itself in the WAN, historically the domain of faster Sonet protocols."
"Working out the details would allow Ethernet, for the first time, to catch up to the speeds of optical transport, specifically the 9.58 Gbit per second OC-192 protocol. "We've never had a point where the Sonet carrier network data rate and the Ethernet data rate were the same," said Paul Bottorff, director of switching architectures at Nortel Networks."
"In addition, the group is mulling several PHY chips, including one that combines four 2.5-Gbit transceivers on a die, aggregating their output via WDM."
"For now, however, Ethernet at any speed remains a curiosity for service providers. That could change if the IEEE chooses a four-channel WDM physical channel, with all four wavelengths operating at 2.5 Gbits per second, as the physical layer for 10-Gbit Ethernet. Interfaces to WAN edge equipment would be eased, and the cost of creating access rings and metropolitan rings for Ethernet would be four to five times lower than the cost of meeting full Sonet jitter specs."
"MAC layer silicon at 0.25 micron and below will be easy to implement in CMOS for 10-Gbit networks. The costs for physical-layer transceivers will depend on how the channel is implemented."
[Note: Noam also mentioned that MRVC has at least 0.25 micron capability with their upgrade.]
"Even serial 10-Gbit transceivers for Ethernet will cost far less than those for OC-192 Sonet, because of Ethernet's relaxed jitter specs; but parallel WDM components may represent the smallest price differential over 1-Gbit transceivers."
In summary, MRVC and NAC are still positioning dead center to future trends, and they should easily be able to be among the first with 10Gbps devices as soon as the preliminary spec becomes clear.
For Metro networks, the price drop due to 10-Gb Ethernet would be a major driving factor. In addition, NAC lowers the DWDM costs much further because they only need 1 wavelength at each node instead of full DWDM devices. The combination should allow NAC to become a big winner in this future market - a fact JNPR and KPCB have not missed.
The full article: techweb.com
|
