When Bob Metcalfe proposed Ethernet to his managers at Xerox PARC in the early 1970s, he unwittingly started a technology phenomenon that is still growing and evolving today.
Ethernet has become the dominant connection technology for most market segments such as enterprise, telecom, and public cloud. In fact, it’s everywhere—you see it in automobiles, trains, beverage dispensers, DVD rental machines, and so on.
The technology is more than 40 years old and while few technologies are still alive after that period of time, Ethernet is still going strong.
From a business perspective, the Ethernet market hit an all time high in revenues and in port shipments in Q3 2014, and is poised to grow to more than $25 billion in sales by 2019, according to industry analysts Dell’Oro Group.
The new industry push towards software-defined networks (SDN) and the evolving deployments of smarter endpoints with the Internet of Things (IoT), are driving requirements for the development of even higher-speed Ethernet standards as well as a new ‘mid-speed’ development that makes Ethernet more ubiquitous and useful than ever.
Ethernet’s Latest Evolution
Several industry consortia have recently formed to develop new standards to bridge the gaps in Ethernet data rates as newer high-speed networking protocols are defined. For example, 2.5GbE and 5GbE are proposals to provide speeds between 1GbE -10GbE; and 25GbE and 50GbE are intermediate speeds to support 100GbE switches.
Intel is the technology and market leader in 10GBASE-T network connectivity. We’ve invested heavily in the technology and are now seeing significant market growth and customer adoption in datacenter markets because the performance is very good over Cat7e copper cabling and the cost is dramatically lower than fiber-optic cable alternatives.
But, there’s a growing demand for higher-speed Ethernet over Cat5e cabling, which is the most widely installed cabling in Fast Ethernet and Gigabit Ethernet access networks.
We spoke to one customer recently who told us about high-performance, network-connected microscopes that need more than a 1GbE pipe, and yet rewiring to support 10GbE is extremely cost prohibitive.
And we’ve also spoken with integrators that support hospitals who have shown us how they can easily fill a 1GbE pipe when transmitting video images from diagnostic machines. But many hospitals are older buildings, similar to universities, where rewiring is even more expensive, making adoption of faster connectivity ‘realistically unachievable’ today.
Much more applicable to IT managers is using 2.5GbE or 5GbE to connect new 802.11AC Wave 2 WiFi access points that are coming to market later this year.
Wave 2 supports speeds exceeding 3Gb/s, which means they need faster uplinks to the network, but in many cases moving to 10GbE means installing new Cat 7a network cabling and that can mean $300 to $1,000 in additional costs per connection on top of the hardware allocation.
Customers are wary of the costs of rewiring their existing buildings, and so they get very excited to hear that 2.5GbE and 5GbE standards will work over Cat 5e copper cabling—making adopting this new technology much easier. With this technology, higher speed networking between bandwidth-constrained endpoints can now become realistically achievable.
Intel is playing an active role in the NBASE-T Alliance, the group that is coordinating the technology development and the hand off of a specification to official IEEE standards organizations. As the market leader in 10GbE, we can help get these solutions to the market quickly.
Once the technology specifications are set, expect products from Intel and others that auto-sense all of the connection speeds from 1GbE to 10GbE.
Ethernet has found a way to change and grow to meet market needs even when faced with more elegant or higher speed competitive technologies. Thanks to the work of the NBASE-T Alliance and other groups working to make Ethernet better, I predict Ethernet will become more useful and continue its ubiquity in the networking market, maybe even for another 40 years.