The Current State and Future of Advanced Optical

The Current State and Future of Advanced Optical Networks July 20, 2004 Rob. Vietzke@uconn. edu Connecticut Education Network University of Connecticut NEREN

Agenda Current State of the Optic Networking n n Metro Regional New Approaches & Hype Broken Reasoning Community Directions 2

Distance scales for U. S. R&E optical networking Distance scale (km) Metro State/ Regional Extended Regional/ National Examples Equipment < 60 UWash USC/ISI(LA), MAX(DC/MD/VA) Dark fiber & end terminals (cwdm, routers, pluggables) < 500 I-WIRE (IL), I-LIGHT (IN), NEREN Optical Amplifiers, filters, etc. > 500 2 nd Tera. Grid Gen Abilene, NLR Add OEO regenerators & O&M $’s 3

CT Ed Net “Metro” Experience Over 1, 200 miles of SMF-28 fiber deployed across Connecticut 160 dark fiber connected K 12, Higher Ed sites n All edge sites are Gig. E w Higher Eds homed to 2 core sites w Up to 4 K 12’s daisy chained between pair of core sites w Some CWDM for metro higher ed rings w Considering 10 G/1 G CWDM combo links 4

CEN Leased Dark Fiber Network 5

Current State: Metro Typical Scenarios for metro optics n Long Haul GBIC, SFP, XFP, or Xenpak w Occasionally POS or ATM long haul on dark fiber n n n Integrated pluggable in router/switch Redundancy through Link State monitoring on multiple paths and L 3 routing Occasionally CWDM or even DWDM added Almost no other statusing other than up/down of optical layer Market leaders could do more … w Make basic power telemetry available from pluggables w Make cross-vendor pluggables usable w Encourage higher-power affordable single-channel applications n By example 10 G-ER+ to match 1 G ZX 6

Current State: Metro CWDM n Passive nature of OADM’s makes rings livable for customer sites in non-C/O environments w Dedicated, redundant Gig. E Paths from site w Maintains link state to core w Reasonable Cost / Performance n n CWDM OADM packaging adds complexity No cost effective test gear/approach for CWDM w Complexity of cabling, patching, testing, cleaning can’t be overstated. n Still arguably a great, cheap, disruptive approach 7

CWDM OADM Options 8

CWDM GBIC & Xen. Pack-ER One Fiber Pair Site B Site A CWDM-MUX-AD-1470 Xenpak-ER West 1470 nm Not Used East 1470 nm 1470 GBIC Router 2λ 1 x GE 1 x 10 GE CWDM-MUX-AD-1470 East West LAMBDA XXXX OADM Xenpak-ER West 1470 nm East 1470 nm Not Used 1470 GBIC Router 9

Current State: Regional 3 rd Generation DWDM Systems n Market competitors moving very much in step with one another w w n n Planning tools Power Management ROADM (Reconfigurable Add/Drop Mux) Other Evolutions Dynamic Optical power management across systems, paths, channels, bands Very manual OADM approaches w Disruptive insertion and balancing w Lots of patch cords within systems & even shelves? ? w Highly touted ROADM typically works with limited wavelength windows on transponders & mux-ponders 10

Current State: Regional 3 rd Gen DWDM (Cont’d) n More Flexible transponders and muxponders w Larger agile frequency range for optical output w Larger variety of interfaces and better stuffing of 2. 5 G and 10 G wrappers n Device Management w Increasingly good telemetry and monitoring of optic power, bit error rate, etc. w May need to know TL-1 or CORBA w Always check on XML & SNMP options? w A whole new OSS to learn and support in many cases n Very Much an analog world w Dispersion, loss, balancing, planning more about gain structure than bits. w Still wonder if a good sound engineer or a physicist would understands these things better than a bit head 11

DWDM OADM Site Block Diagram 12

Current State: Regional Analog World n n 99% of designs are fundamentally about how much “noise” can you allow to grow in a system before you won’t be able to see bits anymore and therefore need to regenerate Faster bit rates are harder to distinguish at shorter distances than slower bit rates 13

Analog-Optical Systems Digital Optical Network Disruptive Thought? OEO everywhere O-E-O O-E-O • Maximize distance between expensive transponders, muxes, etc • Reduce “noise” by reducing OADM’s, amps, etc wherever possible • Aggregate analog PM data, have some fault isolation capability • Regenerate every analog wavelength at every site, make noise problems so far beyond span specs as to become irrelevant • Leverage E in OEO to create electrical add-drop fabric at each site 14

That’s fine, but what would Next Generation Equipment that you wanted look like?

Needs & Future Directions n Agile Amplification & Dispersion Approach w Not visit dependant as channel loading increases or fiber ages w Not even to patch an agile transponder in to the right ROADM port n n n Interruption-free upgrades/changes Initially affordable & predictable scalability Allow multi-point exchange of lambdas w Not just rings! Spurs & Aliens necessary w Support RON interconnection n Good growth cost curve (>? Distruptive ? ) 16

Observations We’re spending lots of dollars on separate software licenses, maintenance contracts and O&M systems for optical and L 3+ services Telemetry still lousy for the large enterprise Am I running a digital optical network or a broadband CATV system n Composite Triple Beat, Carrier to Noise Ratio, Harmonics, Composite Second Order, etc. all things I remember from balancing amplifier cascades. (Telemetry for that stunk too. ) 17

Current Approach: Assume optics and routers are separate n n Accept need to purchase/learn new O&M Look to roll wrappers in to transport layer w Increase complexity there too? n Look for additional flexibility and features to be built in to –both- parts of the equation 18

Approach: OEO w/elec. fabric Still assume optics and routing is separate Essentially eliminate analog portion of the problem by doing full conversion and retiming to digital at each site Continues separation of Optic and Services hardware Provides switched “wavelengths” cheap 2. 5 G based cross fabric may create new barriers for wide-band migration to 40 G and 100 G Another highly complex device to understand, manage, troubleshoot and maintain Idea of a backbone electrical fabric probably increases opportunities for good multi-point junctions at high bandwidth Replace Line-Card router optics with this technology? 19

Approach: Purify, Simplify O & E Move digital timing, framing, shaping, etc. in to routers with controllable ITU grid wavelengths out (Large form pluggables? ) n Maybe even tunable wavelengths? Make optical portion pure optic amplification, Dispersion, balancing, etc. Eliminate shaping, timing, framing cards. n n n Good telemetry and control back to routers for optic control Can I graph optical performance on Cricket/MRTG? Need dispersion compensation for 10 G or e. FEC at day 0 Advance Alien Wavelengths in to Optic platforms n Think about federated optical networks! Cannibalizes vendor business units & requires current business to do development together Single wavelength application can actually be router-only with no throw -aways as DWDM is added Probably still lousy at multi-directional fiber intersections 20

Other thoughts Drive vendors on telemetry from optics Anticipate we will need the ability to link RON’s, perhaps with (or even without) a common national backbone for all services We need to solve Alien Wavelength problem on optical & router platforms n n Need ability to monitor and control Power, Wavelength, dispersion, e. FEC, etc across the common control plane Need standardization of signalling and wrappers for alien wavelengths 21

The Current State and Future of Advanced Optical Networks July 20, 2004 Rob. Vietzke@uconn. edu Connecticut Education Network University of Connecticut NEREN 22