When designing 100G Ethernet links for campus, metro, or regional networks, network engineers face a bewildering array of QSFP28 transceiver options: QSFP28 100G LR4 (10km), QSFP28 100G ER4 (40km), QSFP28 100G ZR4 (80km), QSFP28 100G BIDI 40KM and 80KM, and specialized QSFP28 100G 100KM coherent modules. Choosing the wrong one can lead to link failures, unnecessary cost, or wasted power. This article provides a systematic, side-by-side comparison of these 100G optics, covering optical parameters, link budgets, dispersion tolerance, fiber requirements, and cost per kilometer. We also discuss when to consider upgrading to 400G solutions such as OSFP112-400G-VSR4 (intra-rack) or QSFP56-DD-400G-DR4 (500m) for shorter, high-density links, and how 100G long-haul optics coexist with 400G data center fabrics. By the end, you will have a clear decision matrix for every 100G link scenario.
All these modules share the same QSFP28 form factor, which means they fit into any QSFP28 cage. However, their optical characteristics differ significantly. The table below summarizes key specifications.
| Transceiver | Standard | Max Distance | Fiber/Connector | Tx Power (dBm) | Rx Sens (dBm) | Link Budget (dB) | Typical Power (W) |
|---|---|---|---|---|---|---|---|
| 100G LR4 | IEEE 802.3ba | 10km | Duplex SMF / LC | -4.3 to +4.5 | -10.6 | 6.3 | 3.5–4.0 |
| 100G ER4 | IEEE 802.3bm | 40km | Duplex SMF / LC | 0 to +4.0 | -18.0 (APD) | 18 | 4.5–5.0 |
| 100G ZR4 | 100G ZR4 MSA | 80km | Duplex SMF / LC | -2 to +4.0 | -23.0 | 21 | 5.5–6.5 |
| 100G BIDI 40KM | Proprietary / MSA | 40km | Single SMF / LC | -2 to +3 | -18.0 | 16 | 4.0–4.5 |
| 100G BIDI 80KM | Proprietary / MSA | 80km | Single SMF / LC | 0 to +4 | -22.0 | 22 | 5.5–6.0 |
| 100G 100KM (coherent) | Coherent ZR/ZR+ | 100–200km | Duplex SMF / LC | -3 to +3 | -25 to -28 | 22–25+ | 6.0–8.0 |
Note: QSFP28 100G LR4 uses four CWDM wavelengths (1295–1310nm) with NRZ modulation. ER4 uses the same wavelength grid but with higher-power lasers and APD receivers. ZR4 uses stronger FEC and tighter dispersion tolerance. BIDI variants use two wavelengths (e.g., 1270nm/1330nm) on a single fiber. The 100KM coherent module uses DP-QPSK and DSP for dispersion compensation.
The most straightforward factor is the physical link distance.
0–10km: QSFP28 100G LR4 is the most cost-effective and widely available solution. Do not use ER4 or ZR4; they are overkill and may require attenuators to avoid overloading the receiver.
10–40km: Two choices: 100G ER4 (duplex fiber) or 100G BIDI 40KM (single fiber). If you have a fiber pair available, ER4 is simpler and more standard. If fiber is scarce, BIDI 40KM halves fiber usage. Cost-wise, ER4 modules are typically more expensive than BIDI (~$1,200 vs $450). However, ER4 offers better interoperability.
40–80km: 100G ZR4 (duplex) vs 100G BIDI 80KM (single fiber). ZR4 is more mature and has wider vendor support. BIDI 80KM is ideal for single-fiber scenarios. Note that some BIDI 80KM modules require APC connectors for reflection control.
80–120km: Standard ZR4 may still work with low-loss fiber (0.20 dB/km) and clean splices, but it is not guaranteed. For reliable operation above 80km, choose a coherent 100G 100KM (ZR+). These modules incorporate DSP to handle dispersion, but they consume more power and are more expensive.
Beyond 120km: Coherent modules with external amplification or 400G ZR solutions are required.
Chromatic dispersion accumulates with distance. Standard single-mode fiber (G.652) has a dispersion coefficient of ~17 ps/(nm·km) at 1550nm. However, 100G LR4/ER4/ZR4 operate in the 1310nm band, where dispersion is near zero (actually ~3 ps/(nm·km)), so dispersion is rarely a limiting factor within their specified reaches. For example, at 80km, dispersion for ZR4 is ~400 ps/nm, well within its tolerance (~1600 ps/nm). The real limit for ZR4 is optical signal-to-noise ratio (OSNR) and loss.
For BIDI modules, which often operate at 1270/1330nm, dispersion is slightly higher but still manageable. Coherent 100KM modules operate in the C-band (1550nm) and include DSP that can compensate for thousands of ps/nm, making dispersion irrelevant.
BIDI (Bidirectional) technology consolidates transmit and receive on a single fiber. Use BIDI when:
You have only one dark fiber strand available (common in municipal or older leased fiber).
You want to double capacity on an existing fiber pair: deploy two BIDI links (one on each fiber) for 200G aggregate.
Fiber lease cost is high per strand – the savings of using one strand instead of two can justify BIDI within months.
You are retrofitting a 10G BIDI link; same wavelength plan may be reused.
Avoid BIDI when:
Your fiber plant has many reflections (dirty connectors, mismatched polishes). BIDI is sensitive to reflected light because the transmitter and receiver share the same fiber.
You need long-term vendor interoperability – BIDI implants vary more than standard ZR4.
You plan to use the same fiber for future 400G coherent (400G coherent typically uses duplex).
In a dense chassis with 48 QSFP28 ports, power differences matter.
LR4 (3.5W): 48 × 3.5W = 168W. Acceptable for most switches.
ER4 (4.5W): 216W – still okay but higher heat.
ZR4 (5.5W): 264W – may require active cooling and careful airflow.
BIDI 80KM (5.5W): similar to ZR4.
100KM coherent (7W): 336W – likely exceeds many switch thermal budgets. Deploy coherent modules only on ports with dedicated cooling or in low-density routers.
If your switch is air-cooled and near its power limit, prefer LR4 or ER4 over ZR4 when possible.
Cost per kilometer (CPK) is a useful metric, but it depends on module price and fiber lease costs.
Assume module prices from the third-party market (as of 2026):
100G LR4: $300 → CPK for 10km = $30/km
100G ER4: $1,100 → CPK for 40km = $27.5/km
100G ZR4: $2,200 → CPK for 80km = $27.5/km
100G BIDI 40KM: $450 → CPK = $11.25/km (plus fiber savings)
100G BIDI 80KM: $1,600 → CPK = $20/km
100G 100KM coherent: $3,500 → CPK = $35/km (but covers longer distance)
BIDI 40KM offers the lowest CPK when fiber is owned. However, if fiber is leased at $200 per strand per month, a 10km link using BIDI saves $200/month in lease cost (one strand vs two), making BIDI pay back in<3 months regardless of module price. For owned fiber, ZR4/ER4 are competitive.
When you upgrade a data center core to 400G, the role of 100G long-haul transceivers remains for metro and access. However, you may need to interconnect 400G switches to 100G WAN links. This requires:
A router or switch that supports both 400G (e.g., using QSFP56-DD-400G-DR4 or OSFP112-400G-VSR4) and 100G QSFP28 ports. Do not attempt to plug a ZR4 module into a 400G QSFP-DD port expecting it to work – form factors differ.
For intra-data center links under 100m, replace 100G LR4 with OSFP112-400G-VSR4 to gain 4× bandwidth at lower power per gigabit.
For links between 100m and 500m, replace 100G LR4 aggregated bundles with a single QSFP56-DD-400G-DR4.
The 100G long-haul infrastructure (ZR4, BIDI, coherent) will likely stay in place for many years, as 400G coherent is still expensive and not needed for many metro links.
Follow this decision tree:
Measure link distance precisely – Use OTDR or GIS data.
Check fiber availability – How many dark strands? If only one strand, forced to use BIDI.
Calculate loss budget – Include connectors, splices, patch panels. If loss >18dB for 40km, consider ZR4 or coherent.
Determine if dispersion is a concern – For
<80km, ignore.="" for="">80km, choose coherent.Check switch thermal/power capacity – For high-density chassis, prefer LR4/ER4 over ZR4.
Evaluate interoperability needs – Multi-vendor environments favor LR4/ER4/ZR4 over BIDI.
Compare total 5-year TCO including fiber lease – BIDI often wins when fiber is leased.
Situation: Two fiber strands available, 45km distance, owned fiber (no lease cost).
Option 1: Use 100G ER4 (duplex). Module cost $1,100. No extra hardware. Power 4.5W.
Option 2: Use 100G BIDI 40KM on a single fiber, leaving the other fiber dark for future expansion. Module cost $450 (per pair). Power 4W.
Option 2 is cheaper upfront and reserves a fiber. However, BIDI 40KM at 45km is beyond its nominal 40km reach – some modules can stretch to 45km with margin, but it is not guaranteed. Option 1 is safer. The owner chooses ER4 for reliability.
Different scenario: Leased fiber at $100/strand/month, 45km. Using BIDI saves $100/month = $1,200/year. Even if a ZR4 or ER4 were free, BIDI would be cheaper after a few months. So BIDI wins.
As 400G coherent ZR optics become more affordable (target <$2,000 by 2027–2028), some metro links may bypass 100G entirely. However, the vast installed base of 10G/100G DWDM will keep QSFP28 100G ZR4 and BIDI in production for a decade. New greenfield metro networks should consider investing in 400G-ready open line systems that support both 100G ZR and 400G ZR+ pluggables.
Within data centers, the transition to 400G VSR4 and DR4 is already in full swing, but for campus and metro, 100G LR4/ER4/ZR4/BIDI will remain the workhorses.
Yes, but you must add an optical attenuator (5-10dB) on the receiver to avoid overloading the sensitive APD. Without attenuation, the link may have high bit errors.
There is no single MSA; different vendors have proprietary implementations. However, many follow a common wavelength plan (1270/1330nm or 1310/1490nm). Always buy matched pairs from the same vendor for guaranteed interoperability.
ER4 uses APD (avalanche photodiode) receivers and higher-power DFB lasers, which cost more to manufacture. The longer reach requires tighter optical margins.
Only if the distance is ≤500m and you have parallel MPO cabling. For duplex fiber, you cannot. Also, the remote equipment must support 400G or be broken out to 100G DR1 (not LR4).
Among 100G options, LR4 has the lowest mW/Gb (35). For 400G, OSFP112-400G-VSR4 has 18.75 mW/Gb, which is nearly twice as efficient as LR4.
Yes. BIDI reflections cause more harm than in duplex systems. Clean both the module and fiber jumper with a one-click cleaner. Use APC polish if specified.
With inline EDFAs, coherent 100G can exceed 500km on the 1550nm window. The module’s own DSP can handle >40,000 ps/nm.
Selecting the correct 100G transceiver for a given link is not a one-size-fits-all decision. QSFP28 100G LR4 remains the default for up to 10km. For 10–40km, choose between ER4 (simpler, standard) and BIDI 40KM (fiber saving). For 40–80km, ZR4 offers reliability and interoperability, while BIDI 80KM excels when fiber is scarce or leased. Beyond 80km, coherent 100G 100KM is the only practical solution. For data center internal links, bypass the 100G LR4/ER4 options altogether and go directly to OSFP112-400G-VSR4 (≤100m) or QSFP56-DD-400G-DR4 (≤500m).
Our team offers free consulting to help you navigate this decision. We supply all the transceivers mentioned – LR4, ER4, ZR4, BIDI 40/80KM, 100KM coherent, and 400G VSR4/DR4 – with full testing and compatibility documentation. Contact us with your link distances, fiber counts, and lease costs, and we will provide a detailed transceiver recommendation and TCO analysis within 24 hours.
Headquarter address :Room 1603, Coolpad Building B, North District of Science and Technology Park, Nanshan District, Shenzhen,China.518057