The jump from 10G to 100G isn't just "ten times faster." It's a fundamentally different cabling architecture - different fiber types, different connector formats, different transceiver modules, and different reach limits that depend on choices you make before a single cable gets pulled.
I've watched data center teams buy $40,000 worth of OM3 multimode trunk cables only to discover their 100G QSFP28 optics needed OM4 minimum for the distances they were running. That's a mistake that doesn't happen at 10G, where the margin for error is wider. At 100G, the specs are tighter and the cost of getting it wrong is higher.
This guide covers the real decision points: which fiber type, which connector, which transceiver, and how far each combination actually reaches - with the IEEE standard numbers to back it up.
100G Ethernet: What's Actually on the Wire?
First, a terminology note that matters: the correct designation is 100G (100 Gigabit), not "100GB." "GB" means Gigabytes, a storage unit. "G" or "Gb" means Gigabit, a transmission rate. You'll see "100GB" on some product pages - it's technically wrong, and in spec sheets and RFPs, using the wrong abbreviation signals that someone hasn't done their homework.
100 Gigabit Ethernet (100GbE) is defined by IEEE 802.3ba (2010) and subsequent amendments. Unlike 10GbE, which uses a single laser on a single wavelength, most 100GbE implementations split the signal across multiple lanes - either 4×25G or 10×10G - to achieve aggregate 100G throughput. How those lanes are carried depends entirely on the fiber type and transceiver module.
Fiber Types for 100G: The Reach Table That Actually Matters
This is the table most buyers need but few articles provide - maximum supported reach for each 100G transceiver type on each fiber grade, per IEEE and MSA specifications:
|
Transceiver Module |
Fiber Type |
Wavelength |
Lanes |
Max Reach |
IEEE Standard |
|
QSFP28 100G-SR4 |
OM3 multimode |
850 nm |
4×25G |
70 m |
802.3bm |
|
QSFP28 100G-SR4 |
OM4 multimode |
850 nm |
4×25G |
100 m |
802.3bm |
|
QSFP28 100G-SR4 |
OM5 multimode |
850 nm |
4×25G |
100 m |
802.3bm |
|
QSFP28 100G-SWDM4 |
OM5 multimode |
850–950 nm (4λ) |
4×25G |
100 m (duplex) |
MSA |
|
QSFP28 100G-DR1 |
OS2 single-mode |
1310 nm |
1×100G |
500 m |
802.3cu |
|
QSFP28 100G-FR1 |
OS2 single-mode |
1310 nm |
1×100G |
2 km |
802.3cu |
|
QSFP28 100G-LR4 |
OS2 single-mode |
1295–1310 nm (4λ) |
4×25G |
10 km |
802.3ba |
|
QSFP28 100G-ER4 |
OS2 single-mode |
1295–1310 nm (4λ) |
4×25G |
40 km |
802.3ba |
The takeaway: OM3 only gets you 70 meters at 100G. That's enough for top-of-rack to end-of-row in some data center designs, but not for longer structured cabling runs. OM4 extends to 100 meters. If your cable plant needs to reach farther than that, you're in single-mode territory.
OM3 vs. OM4 vs. OM5: Which Multimode Grade for 100G?
All three are 50 µm core, laser-optimized multimode fiber. The difference is modal bandwidth - the capacity to carry high-speed signals over distance:
|
Fiber Grade |
Modal Bandwidth (850 nm) |
100G SR4 Reach |
Color Code |
Best Use |
|
OM3 |
2,000 MHz·km |
70 m |
Aqua |
Legacy installs, very short runs |
|
OM4 |
4,700 MHz·km |
100 m |
Aqua / Violet |
Standard data center structured cabling |
|
OM5 |
4,700 MHz·km (+ 2,470 @ 953 nm) |
100 m (SR4) / 150 m (SWDM) |
Lime green |
Future-proofing, SWDM applications |
The practical advice: If you're pulling new multimode cable today for a 100G data center, OM4 is the minimum. OM3 is a legacy grade that limits your reach to 70 meters - fine for some deployments, but it paints you into a corner as rack densities increase and cable paths get longer.
OM5 adds wideband capability for SWDM (Shortwave Wavelength Division Multiplexing), which sends 4 wavelengths over a single duplex fiber pair instead of requiring 8 or 12 fibers via MPO breakout. The fiber costs more per meter, but for greenfield builds anticipating 200G/400G migration, OM5's duplex advantage can reduce total cabling cost significantly.
Connector Types: MPO vs. LC at 100G
This is where 100G cabling gets physically different from 10G.
100G-SR4 uses an MPO-12 connector with 8 active fibers (4 transmit, 4 receive) in a single multi-fiber push-on interface. This is a parallel optics approach - each of the 4 lanes gets its own fiber pair.
100G-SWDM4, DR1, FR1, LR4, ER4 use duplex LC connectors - two fibers, one transmit, one receive. The multiplexing happens inside the transceiver (wavelength muxing for SWDM4/LR4, PAM4 modulation for DR1/FR1).
For structured cabling, this distinction drives infrastructure design:
|
Approach |
Connector |
Fiber Count |
Fiber Type |
Trunk Cable |
|
Parallel (SR4) |
MPO-12 |
8 fibers per link |
OM4/OM5 |
MPO trunk + LC breakout cassettes |
|
Duplex (SWDM4/LR4) |
LC duplex |
2 fibers per link |
OM5 or OS2 |
Standard duplex patch cords |
Parallel optics (MPO) dominates current 100G data center deployments because SR4 transceivers are the cheapest 100G optics available. But the infrastructure cost of MPO trunks, cassettes, and patch panels adds up. As SWDM4 and single-lambda 100G transceivers (DR1, FR1) mature and drop in price, the industry is shifting toward duplex for 100G and reserving MPO parallel optics for 400G and 800G.
If you're building or upgrading a structured cabling system, use high-quality fiber optic connectors and patch cords rated for the insertion loss budgets that 100G demands. At 100G, every 0.1 dB matters more than at 10G - the link loss budget is tighter.
The Migration Question: 10G → 40G → 100G → 400G
One of the smartest things you can do when planning 100G infrastructure is design for the next speed tier. Here's how the upgrade path typically works:
10G to 100G on multimode: If your existing plant is OM4, you can reuse the fiber. Replace 10G SFP+ transceivers with 100G QSFP28 SR4, swap LC patch panels for MPO cassette panels, and you're running 100G on the same fiber. OM3 works too, but only to 70 meters.
100G to 400G on multimode: 400G-SR8 uses MPO-16 with 16 fibers. If you installed 24-fiber MPO trunks for 100G (a best practice), you have capacity for 400G without re-cabling.
100G to 400G on single-mode: 400G-DR4 uses the same MPO-12 connector footprint as 100G-SR4 but on OS2 single-mode. If you installed single-mode from the start, the transition is transceiver-only.
The lesson: pull more fiber than you need today. A 24-fiber MPO trunk costs marginally more than a 12-fiber trunk but gives you a 400G migration path without touching the cable plant. For single-mode runs, installing spare fibers in the same conduit costs almost nothing at installation time and saves a fortune later.
Installation Tips from the Field
Test every connection. At 100G, the total link loss budget on SR4 over OM4 is approximately 1.9 dB - that includes cable attenuation, splice loss, and connector insertion loss. With an MPO connector at each end plus a cassette breakout, you have maybe four mated connector pairs in the path. At 0.35 dB per MPO connection (typical), you're already at 1.4 dB before cable loss. There's almost no margin. Every dirty or misaligned connector eats into a budget that's already razor-thin.
Clean MPO connectors religiously. MPO ferrules have 8 or 12 fiber endfaces in a single connector - one contaminated fiber kills the entire link. Use an MPO-specific one-click cleaner and inspect every ferrule with an MPO inspection scope before mating.
Don't mix fiber grades. Running an OM3 patch cord on an OM4 trunk degrades the entire link to OM3 performance. Every component in the channel - trunk cable, patch cords, cassettes - must be the same fiber grade or better.
Label everything. At 100G densities, with MPO trunks, breakout cassettes, and duplex patch cords all in the same cabinet, unlabeled cabling becomes untraceable cabling. Label both ends of every cable, every panel port, every cassette position. Your future self - or the next tech who touches this cabinet - will thank you.
Frequently Asked Questions
Q: Can I Run 100G Over Existing OM3 Fiber?
A: Yes, but only to 70 meters with 100G-SR4 transceivers. If your cable runs exceed that, you'll need OM4 (100 m) or single-mode with LR4/DR1 optics. Measure your actual installed cable lengths before committing to a transceiver type.
Q: What's The Cheapest Way To Deploy 100G In A Data Center?
A: 100G-SR4 QSFP28 transceivers over OM4 fiber with MPO connectivity is currently the lowest cost-per-port for runs under 100 meters. SR4 optics are widely available, competitively priced, and the parallel architecture is well-understood. For longer reaches, 100G-DR1 on OS2 single-mode is increasingly cost-effective.
Q: Is OM5 Worth The Extra Cost?
A: For new builds where you anticipate 200G or 400G within 3–5 years, yes. OM5 enables SWDM transceivers that carry 100G over a simple duplex LC connection instead of an 8-fiber MPO - reducing patch panel density and simplifying cable management. For short-lifecycle or budget-constrained installs, OM4 remains the practical choice.
Q: Do I Need APC Or UPC Connectors For 100G?
A: Most 100G multimode data center applications use UPC connectors (LC UPC, MPO UPC). APC is required for single-mode applications where return loss is critical - DWDM, long-haul, and PON. If your 100G deployment uses single-mode LR4 or ER4 optics, check the transceiver spec sheet for APC requirements. For more on APC vs. UPC selection, see our SC APC connector guide.
Q: Where Can I Source 100G-Rated Fiber Optic Connectors And Patch Cords?
A: Evolux Fiber manufactures fiber optic connectors (LC, SC, MPO), patch cords, and PLC splitters with OEM customization and 100% factory testing. Contact us for volume pricing and custom cable assemblies.






