Modern Optical Transmission Standards and Transceivers: 10G, 25G, 50G, 100G Ethernet
Table of contents
This article continues the series on legacy optical transmission standards and legacy transceivers, moving from older formats to modern high-speed Ethernet. It is written for engineers and network specialists who need to understand the current landscape — from 10G to 100G and beyond. The focus is on physical layer standards, transceiver form factors (SFP, QSFP, CFP families), and practical considerations for deploying these technologies in data centers of different tiers, telecom environments, and enterprise networks
Key features
- Overview of Ethernet optical standards for 10G, 25G, 50G, and 100G, including reach and fiber type specifications
- Detailed explanation of transceiver form factors: SFP+ (10G), SFP28 (25G), SFP56 (50G), SFP112 (100G), QSFP28 (100G), QSFP56 (200G), and CFP variants
- Comparison of modulation techniques: NRZ vs PAM4 and why it matters for higher speeds
- Lane architectures and how they scale bandwidth
- Fiber types and transmission distances for each standard
- Practical guidance on where each form factor is typically used — from colocation deployments and dedicated hosting infrastructure to direct cloud connectivity and virtual PoP architectures
Introduction
Optical transport networks have entered a phase of high-speed innovation, supporting growth from 10 Gbps up to 100 Gbps per interface — and paving the way for even higher rates. From submarine cable infrastructure to internal data center interconnects, modern networks increasingly depend on dense, power-efficient, and scalable optical interfaces. This article explores current standards, fiber types, lane architectures, and the evolution of transceiver form factors, including the latest SFP56, SFP112, and CFP variants that are critical for scalable infrastructure in data centers, telecom, and enterprise networks.
Ethernet Optical Transmission Standards
10 Gigabit Ethernet (10G)
- 10GBASE-SR: Multimode, 850 nm, ~300 m
- 10GBASE-LR: Single-mode, 1310 nm, up to 10 km
- 10GBASE-ER: Single-mode, 1550 nm, up to 40 km
25 Gigabit Ethernet (25G)
- 25GBASE-SR: Multimode, ~100 m at 850 nm
- 25GBASE-LR: Single-mode, 1310 nm, up to 10 km, commonly used in carrier and enterprise scenarios including international connectivity such as IPLC services
50 Gigabit Ethernet (50G)
- 50GBASE-CR/LR/SR: Emerging for backbone, interconnect, and high-density server applications. These standards adopt advanced modulation (PAM4) and support both single-mode and multimode fiber. In practice, such interfaces are especially relevant for Ethernet transport services including Ethernet Private Line (EPL)
100 Gigabit Ethernet (100G)
- 100GBASE-SR4: 4 lanes × 25G, multimode, ~100 m
- 100GBASE-LR4: WDM over single-mode, 4×25G, up to 10 km; a common choice for backbone and aggregation scenarios connected to IP transit infrastructure
- 100GBASE-ER4/CWDM4: Extended or metro reach, single-mode, CWDM multiplexing; these solutions are also relevant in environments using IX transit.
Advanced Transceiver Form Factors
SFP Family
- SFP+ (10G): Small, hot-swappable, up to 10 Gbps
- SFP28 (25G): Same form factor as SFP+, supports 25G lanes
- SFP56 (50G): Next evolution; identical dimensions, but supports PAM4 signaling for true 50 Gbps per lane. Ideal for high-density switch and server uplinks
- SFP112 (100G): Introduced for future single-lane 100G environments, leveraging advanced modulation for massive bandwidth in compact form
QSFP Family
- QSFP+ (40G): 4×10G lanes
- QSFP28 (100G): 4×25G lanes
- QSFP56 (200G): 4×50G lanes, PAM4 support
CFP Family
Large-form-factor modules standardized for long reach and high aggregate bandwidth, mainly in telecom and data center backbone links:
- CFP (C Form-factor Pluggable, up to 100G): 82×13.6×144.75 mm, supports 10×10G lanes or WDM, designed for long-range, high-power applications
- CFP2: ~41×12.4×107.5 mm, smaller, supports higher density, efficient up to 100G and 200G
- CFP4: ~24×12.4×100 mm, even more compact, similar density as CFP2, adopted for short- and long-reach 100G deployments
These CFP modules enable coherent transmission techniques and higher power budgets needed for DWDM, long-haul, and metro networks. In transport architectures, they are often associated with service layers based on MPLS and Ethernet-over-packet transport models such as EoMPLS.
Comparison Table: Key Transceiver Types
| Transceiver | Speed(s) Supported | Physical Size (W×H×L mm) | Lane Count | Modulation | Typical Power (W) | Main Use Cases |
|---|---|---|---|---|---|---|
| SFP+ | 10G | 13.4×8.5×56.5 | 1 | NRZ | 0.8–2.0 | Server uplinks, short/long reach |
| SFP28 | 25G | Same as SFP+ | 1 | NRZ | 1.5–3.5 | Data center, high-density 25G ports |
| SFP56 | 50G | Same as SFP+ | 1 | PAM4 | 2–4 | Switch uplinks, emerging 50G server links |
| SFP112 | 100G (future) | Same as SFP+ | 1 | PAM4 | ~4–8 | Hyperscale cloud, next-gen aggregation |
| QSFP28 | 100G (4×25G) | 18.35×9.5×72.5 | 4 | NRZ | 4–7 | High-density switch, backbone |
| QSFP56 | 200G (4×50G) | Same as QSFP28 | 4 | PAM4 | 5–10 | Data center backbone, aggregation |
| CFP | 40G–100G | 82×13.6×144.75 | up to 10 | NRZ / WDM | ~24 | Telecom long-haul, DWDM, high power |
| CFP2 | 100G–200G | ~41×12.4×107.5 | up to 8 | PAM4 / WDM | ~12 | Metro, backbone, dense deployments |
| CFP4 | 100G | ~24×12.4×100 | up to 4 | NRZ / WDM | ~6 | Compact backbone, dense DWDM |
Fiber Types and Transmission Distances
| Speed | Standard | Fiber Type | Lane Configuration | Wavelength(s) (nm) | Typical Max Distance | Notes |
|---|---|---|---|---|---|---|
| 10 Gbps | 10GBASE-SR | Multimode (OM3/OM4) | Single lane, 10 Gbps | 850 | ~300 m | Short reach, cost-effective |
| 10 Gbps | 10GBASE-LR | Single-mode | Single lane, 10 Gbps | 1310 | 10 km | Long reach, typical in carriers |
| 10 Gbps | 10GBASE-ER | Single-mode | Single lane, 10 Gbps | 1550 | 40 km | Extended reach |
| 25 Gbps | 25GBASE-SR | Multimode (OM4) | Single lane, 25 Gbps | 850 | ~100 m | Used in data centers |
| 25 Gbps | 25GBASE-LR | Single-mode | Single lane, 25 Gbps | 1310 | 10 km | Carrier and enterprise applications |
| 100 Gbps | 100GBASE-SR4 | Multimode (OM4) | 4 lanes × 25 Gbps | 850 (4 wavelengths) | 100–150 m | Parallel fiber; data center focus |
| 100 Gbps | 100GBASE-LR4 | Single-mode | 4 lanes × 25 Gbps WDM | ~1270–1330 | 10 km | Wavelength division multiplexing (WDM) |
| 100 Gbps | 100GBASE-ER4 | Single-mode | 4 lanes × 25 Gbps WDM | ~1575–1615 | 40 km | Extended reach variant |
| 100 Gbps | 100GBASE-CWDM4 | Single-mode | 4 lanes × 25 Gbps CWDM | CWDM wavelengths | ~2 km | Lower cost short reach for metro |
All standards can be implemented in various modules/fiber combinations, with lane architectures and modulation matched to distance and reach needs within the network topology.
Dominance and Use Cases
Why are SFP28, SFP56, SFP112, QSFP, and CFP families so popular?
- Form-factor compatibility: SFP variants allow platforms to upgrade speeds with minimal hardware changes
- Lane scalability: QSFP and CFP architectures aggregate lanes efficiently for ever-larger bandwidth, which is especially important in Internet Exchange Point (IXP) environments
- Power and density: CFP2/CFP4 offer compact, efficient modules critical for metro and backbone deployments, while QSFP28 and SFP112 provide hyperscale density relevant to high-throughput modern network environments.
- Vendor support and standards: All major switch and optics vendors — Cisco, Finisar, Juniper, Arista, Ciena — actively develop, certify, and deploy these formats
Key Takeaways
Modern Ethernet speeds from 10G to 100G are supported by optical standards optimized for different fiber types, distances, and applications — from short-reach multimode inside data centers to long-haul single-mode for carrier networks.
The SFP and QSFP families have evolved to deliver higher speeds while maintaining physical compatibility: SFP+ (10G) → SFP28 (25G) → SFP56 (50G), and QSFP28 (100G) → QSFP56 (200G). This allows platforms to upgrade bandwidth with minimal hardware changes.
Modulation matters: NRZ works up to 25G per lane; PAM4 doubles data rates for 50G and beyond but requires tighter signal integrity. CFP modules remain relevant for long-haul DWDM applications where higher power budgets are acceptable.
Fiber type selection is critical — multimode for short reach inside data centers, single-mode for longer distances between facilities. Matching the transceiver to the fiber determines whether the link works at all.
Form-factor compatibility and lane scalability are why SFP, QSFP, and CFP families dominate across data center, telecom, and enterprise networks: they provide a clear upgrade path without forklift replacements and fit naturally into the broader IPTP Networks products and services ecosystem