SAN JOSE – Data traffic is expected to grow ten-fold from 2010 to 2015 and up to 100-fold by 2020, according to a new report from the IEEE’s Ethernet standards group. The assessment aimed to lay a foundation for future work on Ethernet interfaces beyond today’s 40 and 100 Gbit/second standards.
“The various projections show traffic increasing to a factor of ten above the 2010 value from as early as 2013 to as late as 2018,” the report concluded.
The IEEE 802.3 Ethernet Bandwidth Assessment group spent a year developing its 44-page report now online. It culls data from eleven presentations across a variety of technical and geographic markets.
The assessment did not attempt to come to grips with what is seen as one of the thorniest issues facing engineers—finding a path to a cost-effective, next-generation Ethernet standard.
“Whether or not these projections are realized will depend, among other things, on the ability of the Ethernet community to keep the cost per bit falling with time… in such a way that the exponential rise in traffic does not result in unsupportable costs,” the report said.
“The question of whether this increased traffic is serviced by the introduction of new rates above 100 Gbits/s or by increasing numbers of the existing interfaces depends on the ability of the higher rates to provide a sufficiently cost-effective solution,” it added.
Ethernet proponents have engaged in a lively debate over the past two years about whether a 400 Gbit/s or a Terabit rate should be the next target. Traditionally Ethernet standards have marched forward at 10x increments, but some engineers say the terabit goal could overstress technology, resulting in unacceptable costs.
For now the debate remains open. “It is left to future standards activities to determine the best way to service these application spaces,” the report said.
In the meantime, the assessment gathered a wealth of data about expanding bandwidth demands.
Globally, the number of Internet users is forecast to increase from 1.9 billion users in 2010 to three billion in 2015. The biggest group will be 1.3 billion users in Asia/Pacific excluding Japan, followed by 495 million in the Middle east and Africa.
Nearly 15 billion fixed and mobile networked devices and machine-to-machine connections will be online by 2015, the report estimated. The average fixed broadband in 2015 is expected to rise to 28 Mbits/s up from 7 Mbits/s in 2010.
In 2015, 68 percent of all broadband connections are forecast to be at least 5 Mbits/s, 40 percent will be at least 10 Mbits/s, and 3 percent will be at least 100 Mbits/s.
The report forecasted Internet Protocol traffic will jump four-fold from 20 exabytes a month in 2010 to 80 in 2015. Mobile IP data will grow fastest at a 92 percent compound annual rate, but it will only account for 7.7 percent of the total by 2015.
Wired links will grow 24 percent to become 46.1 percent of the total in 2015. Wi-Fi links will grow faster at 39 percent to nudge past wired links become 46.2 percent of IP traffic, the report said.
Wi-Fi is expected to dominate Internet Protocol traffic.
@iniewski; both camps did! The 4X QSFP with 10Gig/lane form factor works well for 40Gig whereas the 10X CXP is the first step for 100Gig. There is also the 25/28Gig/lane 4X version for 100Gig. These two rates seem to be defining the limit of Copper interconnects in the datacenter for now. Future versions at 1Tb/sec will be all fiber.
Some facing nasty cluster interconnects at data centers like Facebook have screamed for terabit, right now.
Engineers who know the costs have been saying "It may be cheaper to re-architect your data center clusters than pay for us to push the physics of light that far."
It is true that there is going to be increase in the number of persons using internet throughout the globe. But simultaneously there are significant work is being carried out to reduce the actual traffic by the means of compression and technology advancements. It is true that HDTV and Live TV is increasing the demand on the internet speed but at the same time advancements in the streaming technologies will not let the traffic boost beyond limits. Yes increasing speeds and high speed interfaces will be in huge demand but they will be able to cope up with the situation.
In the past few years, Internet in China and India are growing in a fast pace in the last few years. Infrastructure are being build. Most major cities are wired pretty well. In addition, Internet TV and mobile TV are growing exceptionally well in Asia. With the introduction of HDTV content, the demand of bandwidth has crazily grown in the region. I can understand the growth up to 2015. For the demand to grow further, II am skeptical unless there is a higher definition TV standard and the development of better wired infrastructure in another part of the world, e.g. some countries in Africa and Southern America. However, if wireless technology comes into a breakthrough, the demand of higher infrastructure bandwidth may increase.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.