The Shift from 100G to 400G in the Data Center
Data center operators, MSPs, and enterprise IT teams are steadily moving core and spine-leaf links from 100G to 400G to keep pace with east-west traffic growth, storage fabrics, and denser compute nodes. But the transition is not a simple swap: it involves new transceiver form factors, new cabling standards, and compatibility rules that determine what fits in which port and how far a link can run.
This guide breaks down QSFP28 vs QSFP-DD (and where QSFP56 and OSFP fit in), compares DAC, AOC, and optical transceivers for different reach requirements, and covers procurement considerations — including OEM vs compatible/coded optics — so your team can plan upgrades with confidence. For broader sourcing context, see our buying guides hub, and browse current inventory in networking.
Transceiver Form Factor Primer: QSFP28, QSFP56, QSFP-DD, OSFP
Each generation of high-speed Ethernet has its own dominant form factor. Understanding lane count and electrical signaling helps explain why some modules are interchangeable and others are not.
| Form Factor | Typical Speed | Lanes / Signaling | Common Use |
|---|---|---|---|
| QSFP28 | 100G | 4x25G NRZ | 100G server/spine-leaf uplinks, widely deployed installed base |
| QSFP56 | 200G | 4x50G PAM4 | 200G uplinks, often a stepping stone between 100G and 400G |
| QSFP-DD | 400G (also runs 200G) | 8x50G PAM4 ("double density," 8 lanes) | 400G spine/core links; backward compatible with lower-speed QSFP modules |
| OSFP | 400G | 8x50G PAM4 | Alternative 400G form factor, physically different from QSFP-DD; common on some switch platforms and GPU/accelerator NICs |
Because port and module specifications evolve by vendor and generation, always confirm exact electrical and optical specs against the switch or NIC vendor's documentation before ordering. For a deeper look at module coding and interoperability, see our SFP transceiver compatibility guide.
Compatibility Rules: Which Module Fits Which Port
Port compatibility is one of the most common sources of RFQ confusion. The key rules:
- QSFP-DD ports are physically backward compatible with QSFP28 and QSFP+ modules — a lower-speed module will fit and run at its native speed in a QSFP-DD cage.
- The reverse is not true. A QSFP-DD module will not fit into a QSFP28 or QSFP+ port. If your switch only has QSFP28 slots, you are limited to 100G (or lower) optics and cabling.
- OSFP and QSFP-DD are not cross-compatible without a physical adapter, since they use different cage and connector mechanics despite both targeting 400G. Confirm which 400G form factor your switch or NIC actually uses before sourcing optics or cables.
- Breakout compatibility depends on the optic type — covered in the 400G optics section below.
When planning a mixed-speed upgrade (e.g., adding 400G spine switches while leaving 100G leaf/server links in place), QSFP-DD's backward compatibility can simplify a phased migration, since existing QSFP28 optics can often be reused in QSFP-DD ports at 100G during the transition.
Cabling Decision: DAC vs AOC vs Optical Transceivers
Once form factor and port compatibility are settled, the next decision is how to connect ports: direct attach copper (DAC), active optical cable (AOC), or optical transceivers paired with fiber. Each trades off cost, power, and reach differently.
| Type | Typical Reach | Cost / Power | Best Use |
|---|---|---|---|
| DAC (passive) | Up to roughly 3m at 100G/400G speeds (occasionally ~5m); active DAC extends this modestly | Lowest cost, lowest power, no transceiver needed | In-rack, top-of-rack switch to server connections |
| AOC | Roughly tens of meters, commonly up to ~30m, some variants to ~100m | Low power, fixed optical cable, no fiber cleaning required | Intra-row and short inter-rack runs where DAC reach is insufficient |
| Optical transceiver + fiber | Varies by module (hundreds of meters to tens of kilometers) | Highest cost and flexibility; module chosen independently of cable run | Inter-row, inter-pod, spine-to-core, and any building/campus-scale links |
As a rule of thumb: use passive DAC wherever the physical run allows it (lowest cost and power draw per link), move to AOC when the run exceeds DAC reach but stays within a row, and reserve optical transceivers plus fiber for longer runs or where flexibility to re-terminate and re-patch matters. Always verify exact reach and power specifications against the specific vendor datasheet, since figures vary by speed grade and cable gauge.
Choosing 400G Optics by Distance: SR8, DR4, FR4, LR4
For runs where DAC and AOC are not practical, 400G optical transceivers (QSFP-DD or OSFP) are available in several reach-optimized types:
- SR8 — short reach over multimode fiber (MMF) using an MPO connector; suited to short intra-data-center runs.
- DR4 — roughly 500m over single-mode fiber (SMF) using an MPO connector, structured as 4x100G lanes.
- FR4 — roughly 2km over SMF using a duplex LC connector, using WDM to carry 400G over a single fiber pair.
- LR4 — roughly 10km over SMF using a duplex LC connector, also WDM-based.
Breakout capability differs by type. A 400G-DR4 module can typically break out into 4x100G links (useful for connecting a 400G switch port to four 100G server or leaf ports). 400G-SR8 can break out to two or four lower-speed links depending on the specific implementation. FR4 and LR4, being WDM-based single-lane-equivalent optics over duplex LC, generally do not physically break out into multiple lower-speed links. Always confirm breakout support and lane mapping against the specific optic's datasheet before designing a breakout topology.
100G optics follow a similar logic, with families such as SR4, DR, FR, and LR available for QSFP28 depending on distance and fiber type — the same reach-first selection approach applies.
OEM vs Compatible/Coded Optics — and Counterfeit Risk
Buyers generally choose between OEM-branded transceivers and third-party "compatible" (coded) transceivers:
- OEM optics are sourced or branded by the switch vendor and are guaranteed to match that vendor's expected coding and support model.
- Compatible/coded transceivers are built to the same electrical and optical standards but programmed to be recognized by a specific switch vendor's OS. These can significantly reduce cost, but they must be correctly coded for the target platform — some switch OS versions enforce strict vendor-ID checks and will reject or flag unrecognized modules.
Regardless of source, counterfeit and remarked optics are a real risk in the secondary market. Always buy from a supplier that can verify authenticity and provide traceable sourcing. See our guide on how to identify genuine vs counterfeit Cisco transceivers for inspection and verification steps. If you are also evaluating used switch hardware alongside new optics, our guide to buying refurbished Cisco switches safely covers similar due-diligence practices.
Network Design Tips: ToR vs Spine-Leaf, and Breakout for 400G
A few practical design considerations for 100G/400G rollouts:
- Top-of-rack (ToR): Favor DAC for server-to-ToR links where distances are under a few meters — it is the lowest-cost, lowest-power option and avoids optics and fiber management entirely.
- Spine-leaf and core links: These runs typically exceed DAC and often AOC reach, making optical transceivers with appropriate fiber (SR8/DR4 for short intra-DC runs, FR4/LR4 for longer or inter-building runs) the practical choice.
- 400G-to-4x100G breakout: Using DR4 (or supported SR8 configurations) to break a single 400G spine port into four 100G leaf or server links can reduce the number of high-density switch ports needed while reusing existing 100G-capable equipment during a phased migration.
- Mixed-generation environments: Because QSFP-DD ports accept QSFP28 modules, teams can stage a 400G spine upgrade while continuing to run 100G at the edge, simplifying budget and rollout phasing.
Storage and compute density upgrades often accompany networking refreshes — if your team is also evaluating drive form factors for new server builds, see our comparison of NVMe U.2 vs U.3 vs M.2 vs EDSFF form factors.
Procurement Checklist for 100G/400G Upgrades
- Confirm the exact port form factor on your target switches and NICs (QSFP28, QSFP56, QSFP-DD, or OSFP) before ordering any modules or cables.
- Verify whether 400G ports are QSFP-DD or OSFP — they are not interchangeable without an adapter.
- Map each physical run's distance to the appropriate cabling type: DAC for in-rack, AOC for intra-row, and optics plus fiber for longer or more flexible runs.
- For 400G optical links, match the optic type (SR8, DR4, FR4, LR4) to actual distance and fiber type (MMF vs SMF), and confirm breakout support if planning 400G-to-100G fan-out.
- Confirm switch OS coding requirements before purchasing compatible/coded optics, and buy only from suppliers who can verify authenticity.
- Plan quantities with spares in mind — optics and DAC/AOC cables are common failure points, and lead times can vary for specific coded SKUs.
- For larger rollouts, consolidate transceivers, cables, and switches into a single RFQ to streamline vendor coordination and freight; see our guide on bulk IT hardware sourcing for MSPs and resellers.
Request a Quote for Your 100G/400G Rollout
Whether you are staging a phased 100G-to-400G migration, standardizing on QSFP-DD for new spine switches, or sourcing verified DAC, AOC, and optical transceivers for a multi-rack build-out, Alo Tech Solutions supports bulk and DDP-landed pricing on transceivers, cables, and switches. Request a Quote — Bulk/DDP Pricing and our team will help match form factors, reach, and coding to your exact deployment.