News
As hyperscale cloud providers and colocation operators scale their metro and regional data center footprints, the demand for high-capacity, cost-efficient Data Center Interconnect (DCI) has never been greater. Two coherent pluggable form factors now dominate the conversation: 400G ZR and 800G ZR+. Both promise to collapse what once required expensive line-card-based transport systems into a QSFP-DD slot — but choosing the wrong one can leave capacity on the table or inflate per-bit costs for years.
400G ZR (and its extended-reach variant ZR+) established the blueprint: a standards-based, OIF-compliant pluggable that fits inside standard QSFP-DD cages, delivering up to 400 Gbps over ~80 km without external amplification. It uses 16QAM modulation at ~60 Gbaud, achieving roughly 4 bits per symbol — a sweet spot for metro DCI links under 120 km.
800G ZR+ represents the next leap. Running at ~120+ Gbaud with probabilistically-shaped constellations (typically 16QAM-PS or higher-order shaping), these modules double per-wavelength capacity to 800 Gbps while maintaining or modestly extending reach. They also introduce features like continuous tuneability across the full C-band and open ROADM compatibility, making them viable not just for point-to-point DCI but for flexible-grid metro/regional optical networks.
| PARAMETER | 400G ZR / ZR+ | 800G ZR+ |
|---|---|---|
| Maximum Line Rate | 400 Gbps | 800 Gbps (single wavelength) |
| Baud Rate | ~60 GBd | ~120+ GBd |
| Modulation | 16QAM (fixed or PS) | 16QAM-PS / higher-order PS |
| Typical Reach (unamplified) | ~80 to 120 km | ~80 to 120+ km |
| Amplified Reach | ~300 to 450 km | ~450 to 1,200 km (with EDFA/Raman) |
| Form Factor | QSFP-DD / OSFP | QSFP-DD / OSFP |
| Power Consumption | ~16–20 W | ~20–28 W |
| C-Band Tunability | Typically fixed or limited | Full C-band tunable |
| Ecosystem Maturity | Mature (deployed since 2022) | Rapidly maturing (2024+) |
Scenario A — Metro DCI at sub-80 km. A colocation provider connecting two facilities across a city with dark fiber. 400G ZR is the pragmatic choice here: lower module cost, ample headroom on reach, and a mature multi-vendor ecosystem. Unless fiber is severely constrained, doubling capacity with 800G ZR+ adds cost without proportional benefit.
Scenario B — Regional ring exceeding 300 km. A carrier or enterprise linking data centers across state lines, leveraging existing EDFA amplifier huts along the route. 800G ZR+ with full C-band tunability allows multiple wavelengths on a single fiber pair via a DWDM MUX/DEMUX. This is where the economics flip: one 800G carrier replaces two 400G carriers and simplifies wavelength planning, especially when paired with optical-layer processing modules.
Scenario C — Capacity-constrained campus mesh. A hyperscaler with a campus of three buildings and only two fiber pairs between each. 800G ZR+ delivers 3.2 Tbps of coherent capacity across four wavelengths on a single pair — effectively future-proofing the link for AI training traffic bursts without trenching new fiber.
Port density matters. A 32-port QSFP-DD switch at 400G gives you 12.8 Tbps; at 800G, the same switch delivers 25.6 Tbps. If your routing platforms support 800G today, the ZR+ upgrade path compresses your cost-per-bit meaningfully.
On the other hand, power and cooling budgets cannot be ignored. An 800G ZR+ module draws roughly 40–50% more power than its 400G counterpart. In dense leaf-spine deployments, this differential across hundreds of ports can strain rack power distribution. Network architects should model total power draw — not just per-module specs — before standardizing.
Interoperability continues to improve. Multi-vendor plugfests in 2024–2025 showed strong 800G ZR+ compatibility between DSP silicon providers and host platforms. Still, early adopters should validate their specific switch/router combination at target reach before large-scale rollout.
