Communication Products

Optical Transceivers · Telecom Synchronization · Network Infrastructure · Custom Solutions

Home > Products > Optics Selection Guides > Coherent Pluggable Optics: Optics Selection Guide
Optics Selection Guides

Coherent Pluggable Optics: Optics Selection Guide


What a coherent pluggable actually is
A coherent pluggable puts the digital signal processor, the tunable laser and the coherent optics into the same housing as an ordinary transceiver, so a router o

What a coherent pluggable actually is

A coherent pluggable puts the digital signal processor, the tunable laser and the coherent optics into the same housing as an ordinary transceiver, so a router or switch port can drive a DWDM wavelength directly. What once needed a transponder card, a grey client optic and a separate shelf now occupies one cage.

This guide covers the pluggable coherent families in use today: 400ZR, OpenZR+, the ZR+ parts at 400G and 800G, and the CFP2-DCO modules used where that form factor is already in service. Rate families are listed on the 400G and 800G pages, and the product column on the coherent optics page.

Why the pluggable form factor changed the transport plan

Before pluggables, a coherent wavelength needed a transponder with its own processor plus a pair of short grey optics to hand the client signal over. The pluggable removes both: one part terminates the client signal and the line side, so a wavelength costs one module instead of a card, two grey parts and the rack space.

The trade-offs are real. The module has to fit the power and thermal envelope of the router cage, so power per port and airflow are part of the ordering decision. Reach depends on the mode the module is set to and on the line system behind it, which is why one part appears in several reach classes.

Recommended coherent pluggables by application

ModuleRate and modeMedia / wavelengthReach classWhere it usually lands
400G QSFP-DD 400ZR400G single carrierSMF, DWDM C-band tunableMetro DCI class, around 120 kmShort data centre interconnect on a point-to-point span
400G QSFP-DD OpenZR+400G, or lower rate modesSMF, DWDM C-band tunableMetro to regional classLinks that reach beyond the metro ring
400G QSFP-DD or OSFP ZR+400GSMF, DWDM C-band tunableAmplified spans of a few hundred kmRegional runs over an existing line system
800G QSFP-DD or OSFP ZR and ZR+800G, or 2 x 400GSMF, DWDM C-band tunableMetro DCI class upward to regionalHigh capacity links at the port limit
100G and 200G CFP2-DCO100G to 200GSMF, DWDM C-band tunableMetro to long haulShelves and cards built around CFP2
400G CFP2-DCO400GSMF, DWDM C-band tunableMetro to long haulExisting coherent line cards in the same cage
Coherent in a lower rate mode100G or 200G per wavelengthSMF, DWDM C-band tunableLonger spans on the same hardwareWhere reach matters more than the headline rate

The reach classes are the standard ones published for each mode; the working limit is the loss and channel plan of the line carried. Worked examples sit on the 400G ZR DCI selection guide and the 800G ZR and 400G DCO interconnect page.

Line system, grid and launch power

A coherent pluggable does not remove the line system; it removes the transponder in front of it. Amplifiers, ROADMs, mux and demux units and the supervisory channel carry the wavelength, and the module has to match the grid, the channel plan and the launch power expected.

Power is the other half of the plan. A coherent module dissipates far more than a grey optic of the same rate and the heat sits inside the host cage, so the rack thermal budget and airflow have to be checked before the port count is fixed. The CFP2-DCO against QSFP-DD ZR+ comparison sets out where the two form factors differ on these points.

Modes, error correction and interoperability

Coherent parts are set to a mode, and the mode decides both the line rate and the forward error correction profile. Both ends must use the same mode and profile, and the host must support both. A 400ZR port and an OpenZR+ port are not interchangeable even though they look identical, so the mode belongs in the request and not just the rate.

Two vendors at the two ends of a link can interoperate when mode, error correction and grid match, but that is a check to make with the actual parts, not an assumption to carry into a build. Keep the mode per link written down and do not mix modes on one line: a module in the wrong mode may come up and then fail under load.

Grey optics, coherent, or both in one network

Coherent is not always cheaper. Over a short span with spare fibre, a grey LR4 or ER4 class part costs less, uses less power and is simpler to turn up. Coherent wins where fibre is scarce, spans are long, or capacity rides on wavelengths in a line system that already exists. Most networks end up with both, split by route length and available dark fibre.

Coding, monitoring and spares

A coherent module has to be supported by the host port, coded for the platform you name and readable through it, so power, temperature and laser bias can be watched from the router. Optics we supply for branded platforms are compatible parts coded for the platform you name, tested against the port, and not manufactured by the brand of the host. Platforms common here include Mellanox, Arista, Cisco and Juniper. Hold spares per rate and per mode.

How to choose, in order

  1. Fix the span and the fibre available, then decide grey or coherent per route.
  2. Choose the mode from the reach and the line system, not from the headline rate.
  3. Check the grid, the channel plan and the launch power the line expects.
  4. Check the host port power budget, airflow and cage count first.
  5. Match the error correction profile at both ends and state whether the far end differs.
  6. Hold spares per rate and mode, and code every part for the platform you name.

Frequently asked questions

What is the difference between 400ZR and OpenZR+?
Both are 400G coherent modes in a pluggable cage. OpenZR+ adds interoperable lower rate modes, which lets the same hardware reach beyond the metro.

Can a coherent module go into a router port instead of a transponder?
Yes, when the port supports the rate, the mode and the power envelope. This is the usual shape of a data centre interconnect build.

Do we still need a mux and amplifiers?
Yes, unless the link is a point-to-point span with no other traffic. The mux units, amplifiers, ROADMs and supervisory channel still carry the wavelength.

Why does one module appear in several reach classes?
Because the class depends on the mode set and on the line behind it. A lower rate mode on the same hardware reaches further than the full rate mode.

What do you need to quote a coherent link?
The span and measured loss, the line system and channel plan, the rate and mode, the host platform and firmware, and the temperature at each end.

Send the span, the line system, the rate and mode and the host platform at each end, and we confirm the module, the mode, the grid position and the coding for every wavelength. Related reading: coherent optics for the product column, APX-CFP2-DCO-400G-640KM and APX-CFP2-DCO-200G-1300KM for the CFP2-DCO parts, and all optics selection guides.