Few networks transition entirely from 100G to 400G overnight. Most operators face years of mixed environments where QSFP28 100G LR4, ER4, ZR4, BIDI 40KM/80KM, and even QSFP28 100G 100KM coherent modules must coexist with newer 400G optics such as OSFP112-400G-VSR4, QSFP56-DD-400G-VSR4, QSFP56-DD-400G-DR4 (also known as QSFP DD DR4), and QSFP112. Interoperability challenges—different fiber counts, modulation formats, FEC requirements, and connector types—can derail deployments. This article provides a practical roadmap for ensuring seamless interoperability between 100G and 400G transceivers, covering breakout topologies, gearbox functions, media conversion, and vendor-specific compatibility considerations.
The fundamental sources of incompatibility include:
Electrical interface: 100G QSFP28 uses 4×25G NRZ or 4×50G PAM4; 400G QSFP-DD uses 8×50G PAM4; OSFP112 uses 4×112G PAM4.
Optical interface: Parallel fibers (e.g., QSFP56-DD-400G-DR4 with MPO-12) vs duplex LC (e.g., QSFP28 100G LR4).
Modulation: NRZ (most 100G) vs PAM4 (400G and some 100G BIDI).
Wavelength plan: CWDM4 (LR4), LAN-WDM (ER4), or dual-wavelength BIDI.
FEC: 400G requires RS-FEC; 100G NRZ does not.
Understanding these differences is the first step to designing mixed-speed links.
Not all plugs fit into all cages. The table below summarizes physical compatibility:
| Module Type | Cage Required | Backward Compatible With |
|---|---|---|
| QSFP28 (100G) | QSFP28 cage | QSFP+ (40G) with adapter |
| QSFP56-DD (400G) | QSFP-DD cage | QSFP28/QSFP56 (electrically, but not mechanically all) |
| OSFP (400G/800G) | OSFP cage | None (different size) |
| QSFP112 (400G) | QSFP112 cage | None (different pinout) |
Critical note: A QSFP56-DD-400G-DR4 module can physically fit into a QSFP-DD cage, but it will not fit into a legacy QSFP28 cage. Conversely, a QSFP28 100G LR4 can be inserted into a QSFP-DD cage (if the cage is designed for backward compatibility), but the switch must support 100G mode on that port. Many modern QSFP-DD switches allow this.
OSFP112-400G-VSR4 modules require OSFP cages. They cannot be used in QSFP-based switches. When planning mixed deployments, ensure your switch supports both form factors or use separate switch families.
One of the most practical interoperability methods is breakout: a single 400G port feeds four 100G links. However, breakout only works when the optical parameters align.
A passive MPO-12 to 4×LC duplex breakout cable can split the four 100G PAM4 optical lanes of a QSFP56-DD-400G-DR4 into four independent 100G signals. The remote side must use 100G DR1 modules (single-lane 1310nm, PAM4, with RS-FEC). These are less common than LR4. Do not attempt to connect to QSFP28 100G LR4 (4 wavelengths, NRZ) – they are incompatible.
Similarly, an OSFP112-400G-VSR4 can break out to four 100G VSR4 links over MMF, each using 850nm VCSELs. This works with QSFP28 100G SR4 modules (if the SR4 also uses 850nm and PAM4 – note that legacy SR4 is NRZ). Most 100G SR4 modules are NRZ, not PAM4. Therefore, breakout from 400G VSR4 (PAM4) to 100G SR4 (NRZ) requires a gearbox. This is often handled inside the switch if the switch supports PAM4-to-NRZ conversion.
Before deploying breakout, verify:
The switch supports breakout mode (e.g., port breakout 400G-4x100G).
The remote 100G modules match the optical lane characteristics (wavelength, modulation, FEC).
FEC is consistently configured: 400G side uses RS-FEC; 100G side may use no FEC or RS-FEC depending on the module.
When passive breakout is impossible due to incompatible modulation or wavelength, active media converters or gearboxes are required. For example, to connect a QSFP56-DD-400G-DR4 (parallel SMF, PAM4) to a QSFP28 100G LR4 (duplex SMF, NRZ), you need an active converter that:
Accepts 400G DR4 input, retimes and converts to four 100G NRZ signals, and then multiplexes four wavelengths onto a duplex fiber (or outputs four separate LC duplex).
Such devices exist as standalone 1RU chassis or as pluggable “gearbox” modules. They add cost ($1,500–$3,000) and latency (a few microseconds). Use them sparingly.
For QSFP28 100G BIDI 80KM to 400G DR4, no direct path exists; use a 100G BIDI to 100G DR1 converter first, then aggregate.
A common scenario: a data center uses 400G OSFP112-400G-VSR4 for intra-rack links and QSFP56-DD-400G-DR4 for spine-leaf. The same data center connects to a remote site via a 100G metro link using QSFP28 100G ZR4 or BIDI 80KM. How do they interconnect?
At the data center edge, you need a router or switch that supports both 400G and 100G ports. For instance, a spine switch with 400G uplinks can also have 100G QSFP28 ports for WAN connectivity. The 400G internal fabric connects to the 100G WAN port via the switch’s ASIC – no special optics needed. The key is to ensure the switch has both port types. If the switch only has 400G ports, you can use a 400G-to-100G breakout (as above) but must match the optical standard of the WAN link – which is unlikely. Better to keep separate edge devices.
Interoperability is not just optical; it is also electronic. Major switch vendors (Cisco, Arista, Juniper, Huawei) maintain approved vendor lists. A QSFP56-DD-400G-DR4 from a third-party supplier may be rejected by the switch’s firmware. Symptoms: module not recognized, interface stays down, or errors appear in log.
Solutions:
Purchase transceivers pre-coded for your switch brand.
Use switch commands to allow unsupported modules (e.g., service unsupported-transceiver on Cisco, unsupported-transceiver on Arista).
Update switch firmware to the latest version, which often expands compatibility.
For QSFP28 100G 100KM coherent modules, vendor locking is especially strict because coherent DSPs require specific firmware. Always test a sample before bulk ordering.
When a 400G PAM4 link (e.g., OSFP112-400G-VSR4) is connected to a 100G NRZ link via a gearbox, FEC domains are separate. The gearbox terminates the RS-FEC of the 400G side and generates (or passes through) the appropriate FEC for the 100G side. Misconfiguration of FEC on either side can cause massive errors.
Best practice: Use auto-negotiation where supported. For manual configuration, ensure the 400G port has RS-FEC enabled, and the 100G port has FEC disabled (for NRZ) or enabled as per the module’s spec.
A financial services firm had an existing leaf-spine network using QSFP28 100G LR4 on duplex SMF (distances up to 300m). They wanted to introduce 400G spine switches while keeping the 100G leaf switches. The solution:
New spine switches with QSFP-DD cages.
Installed QSFP56-DD-400G-DR4 modules in the spine.
Used active breakout transponders (external 1RU) to convert each 400G DR4 port to four 100G LR4 signals (changing parallel SMF to duplex, and PAM4 to NRZ).
Connected the 100G LR4 outputs to existing leaf switches.
Cost: $2,500 per 400G port for the transponder, plus $1,100 for the DR4 module. They phased out the transponders as leaf switches were upgraded to native 400G. This hybrid approach allowed zero downtime migration.
Although this article focuses on 100G/400G mixing, note that 100G-to-100G interoperability also has pitfalls:
QSFP28 100G LR4 and ER4 can interoperate if the distance is within 10km (LR4 to ER4 works because ER4 is more sensitive).
QSFP28 100G ZR4 and LR4 can interoperate over short distances, but the ZR4’s higher transmit power may overload the LR4 receiver. Use attenuators if needed.
QSFP28 100G BIDI 40KM and BIDI 80KM are not interoperable with each other or with LR4/ZR4 due to different wavelength pairs.
QSFP28 100G 100KM coherent modules are typically not interoperable with direct-detect ZR4.
Always check the optical specifications before connecting different 100G standards.
As 800G emerges, backward compatibility will be limited. OSFP112 will support 800G by using 8 lanes (800G SR8/DR8). Existing OSFP112-400G-VSR4 modules can be used in 800G switches, but the switch will run them at 400G. Conversely, 800G modules will not work in 400G-only OSFP cages if the cage lacks the high-speed signaling. Plan for a similar transition period.
No. DR4 uses 4 parallel fibers and PAM4; LR4 uses duplex fiber and NRZ with 4 wavelengths. They are optically and electrically incompatible. You need an active converter.
Yes, if the QSFP-DD cage is backward compatible with QSFP28 and the switch supports 100G BIDI. Many modern switches do. Check the vendor’s compatibility matrix.
No. OSFP and QSFP-DD are physically different. You would need an OSFP-to-QSFP-DD adapter (rare) which adds complexity. Use the correct form factor for your switch.
Use a 400G DR4 to 4×100G SR4 active breakout cable (sometimes called a “gearbox breakout cable”). It converts the SMF parallel to MMF parallel and adjusts modulation. Passive cables will not work.
No. QSFP112 uses a 4-lane 112G electrical interface; QSFP-DD expects 8 lanes. They are not interchangeable, and no adapter exists.
Usually yes, if both comply with the ZR4 MSA. However, some vendors implement proprietary FEC or dispersion compensation. Test before deployment.
Check the switch’s configuration guide. Look for commands like “breakout” or “port-group”. Many Arista, Cisco Nexus, and Juniper QFX switches support breakout.
Successfully mixing 100G and 400G transceivers—from QSFP28 100G LR4/ER4/ZR4/BIDI/100KM to OSFP112-400G-VSR4, QSFP56-DD-400G-VSR4, QSFP56-DD-400G-DR4, and QSFP112—requires careful planning. Key takeaways:
Match form factors or use breakout cables only when optical parameters align.
Use active gearboxes for incompatible modulation or fiber types.
Be aware of vendor coding and FEC requirements.
Test interoperability in a lab before production deployment.
Our interoperability lab offers pre-deployment testing for any combination of 100G and 400G optics. We stock all the transceivers and breakout cables discussed, and our engineers can provide a certified compatibility report for your specific switch models. Contact us to schedule a compatibility test or to request a custom hybrid deployment design.
Headquarter address :Room 1603, Coolpad Building B, North District of Science and Technology Park, Nanshan District, Shenzhen,China.518057
