Are we seeing the end of 10 Gigabit Ethernet?

The advent of 40GbE and 100GbE are fuelling rumours of the end of 10GbE

Tags: Brocade (www.brocade.com)FVC - First Video Communications IncorporationGartner IncorporationNexansSiemon
  • E-Mail
Are we seeing the end of 10 Gigabit Ethernet? Samer Ismair from Brocade says that the industry will focus on making 40Gb and 100Gb interfaces more economical before moving to the next speed.
By  Piers Ford Published  March 26, 2013

The advent of 40GbE and 100GbE are fuelling rumours of the end of 10GbE; while this rumour may be an exaggeration, the new, higher speed Ethernet standards are opening the door to higher capacity networks, writes Piers Ford.

Rumours of the death of 10 Gigabit Ethernet as the high-speed network standard of choice might be exaggerated. But there’s no doubt that the advent of the Institute of Electrical and Electronics Engineers (IEEE) 40 Gigabit Ethernet (40GbE) standard, and the promise of cost-reducing enhancements to its 100 Gigabit Ethernet (100GbE) standards, are fuelling a sense of anticipation among network managers, particularly those who are charged with developing SANs and datacentres that meet the heavy duty demands of today’s enterprises.

As Samer Ismair, MENA systems engineer at high-performance network specialist Brocade Communications, points out, nobody is asking for networks to be slower. And the idea of a single killer app driving demand for speed in specific situations is a simple misnomer. Everywhere, the hunger for - and expectation of - high-bandwidth, on-demand, personalised content is piling on the pressure to speed up the network, internally and at carrier level.

“These two revolutionary new Ethernet speeds open the door to significantly higher-capacity networks and enable networks to scale in ways that were previously impossible using 10 Gigabit Ethernet,” he says.

“The proliferation of high-bandwidth applications to millions and even billions of users will continue to drive higher bandwidth requirements. The industry will focus on making 40Gb and 100Gb interfaces more economical and available in higher-density line cards first, before jumping to the next speed.

“40GbE is targeted mostly at datacentre core and aggregation solutions such as Tor Server Aggregation, Blade Server Access, Metro Core and Campus Core. 100GbE mostly targets Metro Core, Large Campus Core and Data Centre Core and Aggregation solutions.”

40GbE interest
While 100GbE is still some way short of offering a datacentre-quality proposition, according to some industry watchers, interest in 40GbE – itself still in the early stages of development - is huge. The industry has learned from cost and copper interface-related energy problems associated with the early adoption of 10GbE, which is currently the dominant high-speed standard.

Tarek Helmy, regional director Gulf & Middle East at Nexans Cabling Solutions, says the arrival of the first Ethernet standard over optical fibres specifying speeds of 40Gb and 100Gb was a significant factor in the evolution of 40GbE. Equally important, in the summer of 2012, the IEEE started the development of a new standard to run 40GbE over balanced copper cabling.

“Obviously the main advantage this new 40GbE is going to bring is its increased speed,” explains Helmy. “But it will also come with specific features affecting the design of datacentre cabling infrastructures that users should be aware of. In particular, there are differences in length of cables that can be allowed for.”

Helmy said that because 40GbE for copper cabling is still in its definition phase, it is unclear what the exact distance limitations are going to be. However, it is obvious that the usual 100m channel lengths are not going to be supported by the new standard.

“The distance limitation is surely going to be set below 50m,” says Helmy. “There is a logic to this, as 40GbE over copper will be predominantly used for switch-to-server connections in bandwidth hungry datacentres. The typical length of these links is 35m or less. Developing a protocol that would support longer lengths is going to add unnecessary extra costs.

Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code