---
title: "Grey Optics vs DWDM for Metro: The Point Where Wavelengths Start Saving Money"
type: comparison
target_audience: technical
score: 9/10
---

The transition from grey optics to DWDM pluggables is the most consequential optical infrastructure decision most enterprise network architects and ISP engineers make, and it almost always gets made too late, after the fiber lease costs have already become embarrassing on a P&L. The economics are counterintuitive: you spend more per port on optics to spend dramatically less on transport infrastructure. Understanding where the crossover point sits requires building an actual model rather than relying on rules of thumb.

Grey optics — the industry's informal term for single-wavelength transceivers operating outside the DWDM C-band grid — cover the practical metro range with two common choices. For 2km to 10km, 1310nm LR (IEEE 802.3ae Clause 49 for 10G, Clause 88 for 100G LR4) is the workhorse. For 10km to 40km, 1550nm ER modules based on directly-modulated or electro-absorption lasers handle distances up to 40km in 100G and to 80km in 10G with appropriate optical budget. Compatible 100G LR4 QSFP28 modules (Flexoptix P.10731) run approximately €120-180 each; 100G ER4 (Flexoptix P.10732) cost approximately €280-400 depending on reach variant. These are the cheapest optical transceivers that will cover metro spans. The problem is that each module occupies an independent fiber pair, and fiber in metro areas costs real money.

Dark fiber lease pricing in metro areas varies significantly by market but runs approximately €500-2,500 per fiber pair per month for intra-city spans of 5-15km in European markets. Frankfurt and Amsterdam, where carrier-neutral facilities concentrate, are at the lower end of this range due to competitive fiber market density; secondary markets like Leipzig, Eindhoven, or Salzburg run at the upper end. A network operator with 8 100G circuits between the same two data centers — which is not unusual once you include redundant paths, separate traffic classes, and capacity reserves — is paying for 8 fiber pairs, or €4,000-20,000 per month purely in fiber lease costs for that one A-to-B metro segment.

DWDM changes this arithmetic completely. ITU-T G.694.1 defines the standard DWDM channel grid with 100GHz spacing across the C-band, providing 40 usable channels between 1530nm and 1565nm, or with 50GHz spacing (now standard for 100G and above), 80 channels. A single fiber pair carrying DWDM can multiplex all 80 channels, each carrying 100G or 200G, over one fiber pair. Eighty 100G circuits over one fiber pair replaces 80 fiber pairs. At €1,000/pair/month, that is €80,000/month in fiber cost reduced to €1,000/month — a €79,000/month improvement. The DWDM optics cost for that scenario (80 QSFP28 DWDM modules at each end): approximately €800-1,200 per fixed-wavelength QSFP28 DWDM module from vendors like Lumentum or Flexoptix P.11101, so €64,000-96,000 for 80 modules at one end, paid once. The ROI at even 4 circuits sharing a fiber pair is positive within months.

The specific QSFP28 DWDM form factor comes in two distinct architectures with significantly different costs. Fixed-wavelength DWDM QSFP28 modules are pre-set to a single ITU channel at the factory — channel 33 at 193.1 THz (1550.92nm), for instance — and cannot be retuned without physical replacement. They cost approximately €800-1,500 each from established vendors. Tunable DWDM QSFP28 modules cover the full C-band (nominally channel 1 through 96 on 50GHz grid, though most implementations cover channels 17-61 for 100GHz spacing or channels 17-122 for 50GHz) and can be programmed to any channel via CMIS or SFF-8636 management interface. Tunable modules from Lumentum, Acacia (now Cisco), or available through Flexoptix run approximately €2,000-3,500 each. The inventory advantage of tunable is compelling: one SKU replaces 80 SKUs, which matters enormously for spare management.

The next tier up is CFP2-DCO (Digital Coherent Optics) for distances beyond what direct-detect QSFP28 DWDM can handle. CFP2-DCO modules from vendors like Coherent (formerly II-VI), Lumentum, and Acacia cover 80km+ with coherent detection, PM-QPSK or 16QAM modulation, and onboard DSP for dispersion compensation. These run €3,000-5,000 per module. For 100G-ZR+ in QSFP28 form factor, the OpenZR+ standard (implemented by Inphi Colorz-II, Acacia AC400, and the OpenZR+ MSA modules) achieves 120km with coherent DP-QPSK, fitting in a standard QSFP28 cage. These represent the current price-performance boundary for metro coherent: approximately €1,500-2,500 per module, 120km reach without external amplification, and QSFP28 form factor that fits existing switch hardware.

The ROI model needs to account for four specific financial variables: fiber lease cost per pair per month, number of parallel A-to-B circuits, distance (which determines whether direct-detect DWDM or coherent is needed), and the amortization period for optics investment. For a network with fewer than 4 parallel circuits between any given pair of sites at fiber lease costs below €800/pair/month, grey optics with multiple fiber pairs is usually cheaper over a 3-year horizon. Above 6 circuits, or when fiber lease cost exceeds €1,200/pair/month, DWDM pays back in under 18 months at 100G rates. The specific inflection point also shifts when rack space is constrained: a 48-port QSFP28 chassis running DWDM carries 48x100G over 2 fibers, while the same chassis with grey optics requires 48 fiber pairs terminated into patch panels that may consume 2-4U of patch panel space alone.

There is a practical distance limitation on direct-detect DWDM QSFP28 that surprises engineers migrating from DWDM line systems: without inline amplification, chromatic dispersion limits 100G NRZ DWDM to approximately 80km on standard SMF-28 (D = 17 ps/nm/km at 1550nm), and without integrated DCM (Dispersion Compensating Module), the accumulated dispersion at 80km is approximately 1,360 ps/nm, which is within direct-detect QSFP28 tolerance only with DSP-based EDC. The coherent QSFP28 ZR and ZR+ modules handle this via the DSP, but conventional direct-detect DWDM QSFP28 modules must operate within their specified reach. A 100G DWDM QSFP28 rated to "80km" on the data sheet means 80km with the specific dispersion budget they tested — span loss and dispersion must both be within spec. A circuit with 60km distance but aging fiber showing 0.35dB/km loss plus high PMD from repeated cable repairs may fall outside the module's budget even at shorter distance.

The organizational reality in most ISPs is that the DWDM transition happens piecemeal: one high-traffic corridor migrates first, then successive rollouts as lease renewals come up for other corridors. For network teams running this transition, Flexoptix provides one tangible operational advantage: they can program and test channel-specific DWDM modules against the customer's target platform before shipment, verifying not only that the wavelength is correct but that the EEPROM configuration will be recognized correctly on the specific NOS version in use. Ordering pre-programmed channel modules from a grey-market vendor that ships generic factory stock means you may receive a module that DOM reports correctly on a lab Arista but behaves differently on a Nokia 7250 IXR where the CMIS driver expects specific OIF field values. The fiber lease savings are too large to risk on untested optics.