Fiber or Copper: What Is The Future Of Ethernet?

How will Ethernet evolve in the future and what needs to be done!

Photo source: Flickr 

Development of a 25 gigabits per second (Gbps) Ethernet may seem retrograde because 40- and 100-Gigabit Ethernets already operate, but its development relates to a need for speed for cloud data center servers, which exchange enormous data. Google and Microsoft promote industry standards for 25-Gbps and 50-Gbps Ethernets, the Institute of Electrical and Electronics Engineers (IEEE) has started a 25-Gbps group, and consensus is developing within the industry. IEEE also has begun work on 50- and 400- Gbps Ethernet standards.

Important to 25-Gigabit Ethernet is the fact that many of its components already exist. The 100-Gigabit standard has four 25-Gbps pathways, and the existence of so many parts should drive development costs down. A 25-Gbps standard also allows network architects to build data centers with servers linked to 25-Gbps switches and 100-Gigabit uplinks, a four-to-one ratio to which they are accustomed.

Modules of 25 GBPS will be important as cloud data centers age to let network engineers reuse this technology as needs grow and speeds increase. As the cloud matures into a real data center business mode, the convenience of reuse will become apparent.

What Users Want

Current laptops can handle perhaps a gigabit per second, assuming they still have Ethernet ports, which the trend to thin laptops has discouraged. Many users probably don't know how fast their Ethernet connections are because their concerns are with broadband Internet, not local network, connections.

But many enterprises, Facebook and Google among them, need ongoing growth in local network capacity. A pressing question for them is whether 400 Gigabits or one Terabit, two and a half times more, per second is the more practical approach. Builders of network hardware tend to favor 400 Gbps, but their potential customers say they prefer the faster option.

Either probably will bundle multiple connection modules into one high-speed virtual connection, but there are cost constraints to bundling, for each link needs its own components, and such redundancy is costly, especially with numerous strands of thick, unwieldy, and heavy copper connections.

The Ethernet port on a personal computer uses connectors with copper cables, but for 100-Gbps Ethernet both copper and fiber-optic are in use. Copper is cheaper but fiber optics reach farther. With 40- and 100-Gbps standards, copper cables can be up to five meters long, raising some practical concerns over how fast the next Ethernet generation should be.

Copper versus Fiber-Optic Connectivity

Copper telephone wire is more than a century old, as is the actual way a phone operates. Telephone wiring goes to a switch, which digitizes the analog voice signal and transmits it on a fiber optic cable. Copper telephone wire can provide 3,000 cycles per second of bandwidth, adequate for voice signals, but then an analog modulator/demodulator must boost bit rates for reasonably rapid transmission and then reduce them for reception at the terminus.

So most phone systems now use mostly fiber-optic cable beyond the subscriber links. Fiber provides 1,000 times as much bandwidth over 100 times longer distances. That higher capacity is costly but even at 1000 times as much for two fibers for transmission and reception still much less than the cost of copper per voice channel, and a fiber-optic cable is a fraction of the size and weight of a copper cable, an important point in system conduits, where fiber cable costs are typically a small fraction of those of copper.

For individual subscriber links the costs are different. A copper connection costs less than $100; a fiber connection would cost ten times as much and require power for transceivers. So fiber optic cables to subscribers are nonstarters except in rural areas where their lines are long. There the longer distance capability of fiber makes it more economical.

Most discussions of copper versus fiber are about local area networks (LANs). LAN wire is in fact more recent than fiber optics. Fiber use goes back over 20 years, but computer networks on unshielded-twisted-pair cable have been in use about 15 years through at least five generations to meet increasing LAN bandwidth needs.

Differentiators for Comparisons

So what should an Ethernet cable be, copper or fiber? Major differentiators are maximum cable length and data transmission rates. For copper cable, maximum effective length is 100 meters, for fiber 2,000 meters if multi-mode and more if single-mode. Data transmission rates for copper Category 5e cable is 100 megabits per second, for Category 6 one Gbps, and for fiber 100 Gbps.

With long cable lengths and high data volume fiber cable may be the clear choice, and the question becomes which specific fiber cable. Alternatively, with short cable runs and data volume within copper cabling capacity, copper makes sense and again the question is which copper cable is best for the application.