Enter for the full results page · Esc to close
Four 100G options separated by fibre type, fibre count and how far you actually need to go.
100G is the most widely deployed high-speed interface in the world, and it comes in four common flavours that are not interchangeable. Two are parallel and need eight fibres; two multiplex onto a duplex pair. One runs on multimode, three need single-mode.
The decision is almost always made by what fibre you already have, not by what the modules cost.
Four parallel lanes at 850 nm on multimode fibre, MPO-12 with eight positions used, 70 metres on OM3 and 100 metres on OM4.
The cheapest 100G module by some margin, because 850 nm VCSELs are the cheapest laser in volume production. If your plant is multimode and the distances are short, nothing beats it on price.
Four parallel lanes at 1310 nm on single-mode fibre, MPO-12 with eight positions used, 500 metres.
An MSA rather than an IEEE standard. It is the single-mode answer to SR4: same parallel structure, same connector, five times the reach because single-mode fibre does not suffer modal dispersion. Where you have MPO single-mode plant and need more than 100 metres, PSM4 is the natural step.
Four CWDM wavelengths from 1271 to 1331 nm on a single-mode duplex pair, LC connector, 2 kilometres.
Also an MSA. It exists specifically to be a cheaper LR4: uncooled lasers on the coarse CWDM grid instead of cooled LAN-WDM ones, trading 10 km of reach for 2 km and a much lower price. For most campus links that trade is obviously correct.
Four LAN-WDM wavelengths around 1300 nm on a duplex pair, LC, 10 kilometres.
The IEEE standard for 10 km 100G, with cooled lasers. More expensive than CWDM4 and the right answer whenever the span genuinely exceeds 2 km.
The fibre type row eliminates half the options immediately. After that it is fibre count against reach.
| 100GBASE-SR4 | 100G PSM4 | 100GBASE-CWDM4 | 100GBASE-LR4 | |
|---|---|---|---|---|
| Fibre type | Multimode OM3/OM4 | Single-mode | Single-mode | Single-mode |
| Reach | 70 m OM3 / 100 m OM4 | 500 m | 2 km | 10 km |
| Wavelength | 850 nm × 4 | 1310 nm × 4 | CWDM 1271–1331 nm | LAN-WDM ~1300 nm |
| Connector | MPO-12 UPC | MPO-12 APC | Duplex LC | Duplex LC |
| Fibres per link | 8 | 8 | 2 | 2 |
| Channel loss budget | 1.9 dB (OM4) | 3.3 dB | 5 dB | 6.3 dB |
| Laser | VCSEL, uncooled | DFB, uncooled | DFB, uncooled | DFB/EML, cooled |
| Standard | IEEE 802.3bm | MSA | MSA | IEEE 802.3ba |
| Relative cost | Lowest | Low | Medium | Highest |
Two practical warnings. First, SR4's 1.9 dB channel budget on OM4 is genuinely tight — four mated pairs at 0.5 dB each is 2.0 dB and you have already failed before the fibre is counted. Check it in the reach lookup with your real connector count, not with two.
Second, PSM4 and SR4 use the same MPO-12 connector but different polish — APC for PSM4, UPC for SR4 — and different fibre entirely. They will physically mate and the link will not work. Colour coding exists for a reason; see APC vs UPC.
Under 100 metres on OM4, SR4 is the cheapest way to move 100 gigabits and it is not close. The constraint is the loss budget rather than the distance — keep the mated pair count down, land the trunk on a harness rather than through two cassettes if you can, and inspect every ferrule.
PSM4 keeps the parallel structure and the MPO connector while single-mode fibre removes the modal dispersion limit. It also breaks out cleanly to four 25G links, which SR4 does too but CWDM4 and LR4 cannot. Remember the polish: APC, not UPC.
Two fibres instead of eight is the whole argument, and it is a strong one when the fibre has to be installed or leased. CWDM4 gives you that at close to PSM4 pricing because the lasers stay uncooled. For most building-to-building links on a campus, this is the right default.
The reach comes from cooled LAN-WDM lasers and you pay for them. Before committing, check whether 2 km would actually do — a surprising number of links specified as LR4 are under a kilometre, and the difference across a large deployment is substantial.
No. They use different fibre — multimode against single-mode — and different polish. The MPO connector is the same shape, which is exactly why this mistake happens. Label the plant clearly.
No. Different wavelength plans entirely: CWDM4 is on the 20 nm coarse grid, LR4 is on tightly spaced LAN-WDM. Both ends of a link must match.
The parallel ones: SR4 and PSM4. Their four lanes are on separate fibres, so a conversion harness can point them at four different devices. CWDM4 and LR4 cannot — separating their lanes needs an optical demultiplexer.
100G single-lambda variants exist — one 100G PAM4 lane rather than four 25G NRZ lanes — and they are attractive because they use two fibres and one laser. They are a different generation of silicon and are not interoperable with SR4, PSM4 or CWDM4.
Yes, to 70 metres rather than 100. If your links are under 70 metres it makes no difference. Check the exact figures for your standard and grade in the reach lookup.
The full 100G range in QSFP28, coded for your switch platform and traffic-tested before dispatch.
Same 400 gigabits, three different reaches — and three completely different cabling bills.
Read the comparison →Fibre Plant & ConnectorsA nine-micron core or a fifty-micron one — and why the cheaper fibre is often the more expensive decision.
Read the comparison →Fibre Plant & ConnectorsTwo of these differ by distance. The third differs by something else entirely — and it is widely mis-sold.
Read the comparison →Fibre Plant & ConnectorsLegacy plant was built in twelves. Modern optics come in eights and sixteens. The arithmetic does not work.
Read the comparison →Send the specification and an engineer answers it — no obligation, and no sales sequence. If the answer is that you do not need what we sell, that is the answer you get.
Usually answered within one working day.