Transcript The Path to 100 G Ethernet

The Path to 100 G Ethernet
Martin Nuss
VP & Chief Technologist
© Ciena Corporation
A need for speed: 10GbE, 40Gb, 100GbE
Confidence:
We will fill up the
bandwidth just like we
fill-up disk space and
memory
© Ciena Corporation
2
40G Transmission – Current Drivers
40G IP Router interfaces
Problems with link aggregation
Most flows small, distribute nicely
Utilization
Link bandwidth requires N X 10G
Large flows from MPLS/IPSEC problematic
Aggregation Group Member
32-40 Ch DWDM ring exhaust
Multi-access DWDM rings in metro
Typically 5 nodes, some larger
At exhaust build entire new ring
New fiber, amps, ROADMs, installation, space, power,…
Or add a pair of 40G transponders
Both of these applications support 40G transponder costs > 4X 10G transponder cost
© Ciena Corporation
3
Mix-n-Match of 10G/40G/100G on Same Fiber System
Channel 1
..
.
Channel 40/80
Mix & Match 10G, 40G and
100G waves on a fiber as
needed using similar
engineering rules
10G
40G
100G (future)
Low revenue per bit for data will not justify new network overbuilds
© Ciena Corporation
4
Increased Spectral Efficiency, More Capacity
Spectral Efficiency
1.2
b/s/Hz
1
0.8
0.6
Higher spectral efficiency
at 100G theoretically
possible
0.4
0.2
0
2.5G
10G
40G
100G
Bit Rate
Spectral efficiency: more bits, same fiber system, no forklift upgrade
© Ciena Corporation
5
Transport technology readiness
Possible technology choices for 100G transmission
Parallel options suitable for 100G on dedicated fiber, limited distances
4 X 25G and 10X10G have been proposed
4x 25G VCSEL WDM likely candidate for client-side optics (achievable with CMOS)
Negative service provider reaction to parallel networking solutions in MAN/WAN
Non-scalability of capacity: multiple waves to manage, ROADM port exhaust
Serial options – new technology to improve propagation, spectral efficiency
Trading speed for complexity – starting at 112 Gb/sec
Polarization multiplexing – divide by 2
Each polarization carries a 56 Gb/s signal
Phase coding – e.g.: Four phase states
Four phases encode two info bits
Symbol rate cut in half to 28 Gbaud/sec
00
© Ciena Corporation
6
01
10
11
OTN WAN mapping topics
100G OTN (OTU4/ODU4/OPU4) must support mapping of 100GbE
100GbE line rate of 103.125Gbps (results in OTU4 rate of ~112Gbps w/ FEC)
Full transparency required (don’t repeat the mistakes made at 10GbE)
100G OTN must support muxing of ODU1/2/3 to new ODU4 layer
OTU4 line rate above 112Gbps is probably not realizable economically
with current technology (means ODU4 support 3xODU3 or 4xODU3 is
probably not realizable)
Ciena has proposed to ITU a ~112Gbps OTU4 line rate that supports
transparent mappings of 100GbE and muxing of ODU1/2e/3e (allowing
transparent mappings of 10GbE and 40GbE) and is moving forward
with development of this rate
© Ciena Corporation
7
Economic readiness
Customers would like
40G transponders at 2.5X 10G transponder cost
100G transponders at similar proportional savings
Business case will initially be made based on:
Economic benefit at the overall network solution level
Including CAPEX and OPEX for IP, DWDM and fiber
A reasonable 100G initial economic target: 100G = 2.5 X cost of 40G
Parity in cost/bit, with improved spectral efficiency
Definition of “acceptable” cost ratio with respect to 10G will vary with customer
© Ciena Corporation
8
Thank You
© Ciena Corporation