400G ZR vs ZR+: You're Asking the Wrong Question
The ZR vs ZR+ debate dominates every WAN discussion. It's the wrong frame. The real question is whether you understand your link budget well enough to buy either.
Every RFP for WAN optics asks the same question: ZR or ZR+? The answer is always the same: it depends. That's not a cop-out. It's the only honest answer, because the question itself is incomplete.
400G ZR (OpenZR, ITU-T G.698.4) targets single-span unamplified links up to 80km. At 100 Gbd baud rate, DP-16QAM, it delivers 400G in a QSFP-DD form factor. Power consumption: 14–16W. Cost: around $2,000 per unit depending on vendor and volume.
400G ZR+ (OpenZR+, OpenROADM compatible) runs higher FEC overhead and supports adaptive modulation down to DP-QPSK for reach extension. Same QSFP-DD housing, same electrical interface. Reach: 120km to 1000km depending on modulation and amplifier spacing. Cost: $3,500–6,000.
The delta looks obvious until you open your actual link budget and find three things nobody puts in slide decks.
That last line is where projects die. ZR's required OSNR of 22 dB assumes a clean fiber plant. Real-world plants with aging splices, contaminated connectors, and 15-year-old cable deliver 17–18 dB. ZR fails. ZR+ with adaptive modulation drops to DP-QPSK and makes it. You weren't asking the wrong module question. You were asking the wrong fiber question.
ZR works when: the span is under 60km, you have freshly characterized fiber, OTDR traces show less than 3 dB total span loss, and you run it on a direct-connect P2P link without amplification. Metro DCI between owned datacenters on known fiber — that's the use case. It's a good use case.
ZR fails when: you lease dark fiber from a carrier and haven't tested it yourself, you need to traverse one amplifier site, or your operations team can't distinguish a contaminated connector from a bad module. ZR's narrow OSNR margin tolerates no guessing.
ZR+ earns its cost delta on amplified links, multi-span WAN circuits, and any scenario where the fiber plant is unknown or aging. The adaptive modulation is the feature — it doesn't just extend reach, it survives degradation. A fiber that delivers 200G one day and 400G the next after a splice repair isn't a problem with ZR+. It's a data point.
For carrier-leased circuits, ZR+ is almost always right unless you've personally tested the fiber with calibrated equipment and have contractual SLAs on OSNR performance. Carriers don't guarantee optical layer parameters. They guarantee bit error rate at the handoff.
One number changes the economics entirely: mean time to repair. ZR modules at $2,000 with 16 spares costs $32,000. ZR+ at $5,000 for the same spare pool: $80,000. If ZR's MTTR is 4 hours (wrong spare, shipped overnight) versus 45 minutes with ZR+ (adaptive modulation survived the incident, no hardware swap), the math favors ZR+ wherever downtime has a real cost.
Measure the fiber. Run the link budget. Then pick the module that matches the margin you actually have — not the one you wish you had.