Data Center Optical Modules: The Form Factor Evolution from 10G to 800G

One faceplate, increasing bandwidth

Every pluggable form factor in a data center is an answer to the same constraint: the physical width of a switch faceplate. The industry's solution has been consistent — keep the module size roughly the same while multiplying the number of electrical lanes inside it and pushing the data rate per lane higher.

The ladder

| Form factor | Typical speeds | Lanes | Where it still fits | |---|---|---|---| | SFP | 155M – 2.5G | 1 | Legacy transmission, telecom access | | SFP+ | 10G | 1 | Server access, 10G uplinks | | SFP28 | 25G | 1 | 25G server access, leaf-spine | | QSFP+ | 40G | 4 | Legacy aggregation | | QSFP28 | 100G | 4 | 100G uplinks, breakout to 4×25G | | QSFP-DD | 200G / 400G | 8 | 400G fabrics, DCI |

QSFP-DD — "double density" — keeps the QSFP footprint but adds a second row of contacts, doubling the lane count to eight. That is what makes 400G possible in a port that once carried 40G.

Optics do not retire on schedule

A common planning mistake is assuming older form factors disappear. They do not. 10G SFP+ remains in wide production because there is a large installed base of switches that only take 10G, and because 10G links are still entirely adequate for many access and management paths. The same logic keeps 40G QSFP+ alive well past its supposed end of life.

Practical consequence: when you plan a site's optics, plan by port, not by trend. Match the module to the equipment you own.

What to check before upgrading a form factor

1. Does the switch line card actually support the higher speed at that port?
2. Does the existing fiber support the reach? A 10G multimode link does not automatically become a 400G multimode link.
3. What is the power and cooling delta across the whole chassis?
4. Is the breakout topology you need supported?

Canjin Technology carries all six form factors, from SFP through QSFP-DD, so a single supplier can cover both legacy replacement and new builds.