High-speed networking: Faster and faster

This year we`ll celebrate the 20th anniversary of Bob Metcalfe`s brilliant invention - Ethernet - the technology that made modern campus networking practical. His technology has continued to evolve and grow. Now with Gigabit and 10 Gigabit Ethernet speeds common, we need to look further ahead to new horizons.

Graham Vorster, chief technology officer at Duxbury Networking, examines the latest incarnation of Metcalfe`s dream: Terabit Ethernet. Is this the ultimate high-speed platform? What will drive its adoption?

An interesting "letter to the editor" appeared on one of the networking Web sites I monitor. The author was complaining about networking vendors who are "pushing" technologies as "must haves" when in fact they are "nice to haves". He was referring to 10 Gigabit Ethernet, which, in his opinion, was not capable of being fully utilised at enterprise level.

While this is not necessarily true, I understand where he is coming from. However, the only logical answer is that vendors are constantly trying to "push the performance envelope" and their R&D departments are working overtime on technologies designed to speed the flow of data around the network.

For example, it was not long ago - 18 months perhaps - that the goal of 10 Gigabit Ethernet seemed far from being realised. The same was true of Gigabit Ethernet to the desktop. Today both are real and available. In fact, vendors such as Dell offer the latter technology as standard equipment on their PCs.

In November 2002 at a San Diego conference hosted by the National Science Foundation, several 10 Gigabit Ethernet case studies were presented highlighting the growing acceptance of the technology in everything from the data centre and the LAN to the MAN and even the WAN.

While the leap from Fast Ethernet (100Mbps) to Gigabit Ethernet (1 000Mbps) was relatively simple, the leap to Terabit Ethernet (1 000 000Mbps) is more than just an issue of wire speed.

Terabit Ethernet (a single link communicating at terabit speeds) exists and has been proven in tests. We have seen a wave of discussions around the challenges posed to designers of switches and routers that have to perform at the terabit level.

Current `"state-of-the-art" centres on multiple Gigabit Switch implementations bundled together, mimicking those common in the carrier switching world.

There are very few switches on the market designed to support more than a few hundred Gigabit Ethernet pipes. There are even fewer that reach up into the multiple hundreds of Gigabit Ethernet ports.

The big question is: how much demand is there for Terabit Ethernet and what will the key drivers be for this technology?

It is estimated that more than 50 million theoretical concurrent telephone calls could be routed down a single strand of fibre optic cable at Terabit Ethernet speeds.

An impressive statistic. However, it is very unlikely that there will ever be an application such as this in the corporate environment.

It is important to remember that "voice" alone does not require significant bandwidth. But how about some of the newer technologies that developers are pointing to for the future?

One such bandwidth-hungry technology is augmented reality (AR). A close cousin of the 1990`s virtual reality, it goes significantly further.

The idea behind virtual reality was to immerse a user inside a virtual, computer-generated world. Although many technologies were used to achieve this impressive effect, the user was always cut off from any view of the real world outside.

Now, with AR, the user`s view of the real world is enhanced with some additional, virtual information.

AR is a combination of a "real" scene and a "virtual" scene generated by the computer.

In essence, it is an augmentation of human perception with virtual objects superimposed on real ones. Ideally, the real and virtual objects will coexist in such a way that the user is not able to distinguish between the real world and the virtual augmentation of it.

Holographic reproduction - which required huge bandwidth to manage the many millions, if not billions - of reference points in a moving, dynamic three-dimensional image, will be used to great effect in the application of this technology.

While many in the adult entertainment field are impatient for the arrival of AR, it will probably find its most socially acceptable niche in medicine.

Many of the medical applications dealing with image-guided surgery will be made with virtual and AR technologies in future. Pre-operative 3D imaging studies of the patient will provide the surgeon with the necessary views of the internal anatomy.

From these images the surgery will be planned. Visualisation of the path through the anatomy to the affected area where, for example, a tumour must be removed, will be done by first creating a 3D model from the multiple views and slices in the preoperative study.

Then a "dry run" of the operation could be performed with an AR-enhanced patient interacting with the surgeon. In such a scenario, the artificial image of the patient will be superimposed on the real-world operating table in a "live" operating theatre.

The surgeon and the medical team involved in the scene would see and be able to interact with "virtual patient" in just the same way as they would with a real one - down to the finest detail, such as blood flowing from the scalpel`s incision.

With this in mind, many industry watchers agree that it won`t be long before devices are released that are capable of actually optimising a terabit inter-switch link.

Already the promotion of "application-aware" Terabit Ethernet switch/routers is evident. Certain vendors are targeting these products at network service providers for future applications in what is being described as the new breed of OPEN MANs (optical Ethernet networks - metropolitan area networks).

The traffic over these MANs will be characterised by media-rich content and time-sensitive applications.

Will there be problems and challenges associated with their implementation? Undoubtedly.

While Terabit Ethernet could well be suited for moving massive amounts of data over mesh-oriented networks, there may be complexities associated with grooming and transporting bundled voice and latency-sensitive data services on WANs.

Until there are more clearly defined standards in place for network-wide quality-of-service (QOS) provisioning and better control over jitter and delay, Terabit Ethernet will most likely remain in the domain of core data networks.

Increasingly, networking equipment vendors agree that Ethernet is the most appropriate platform for voice, video, storage, security, management and other so-called new generation multi-media applications.

Currently, 10 Gigabit Ethernet is proposed as the best aggregation technology for Gigabit Ethernet desktop connections.

While not too many organisations - given the budget constraints of the economy - are installing such ultra-high-end technology, there is nevertheless a great deal of work going on behind the scenes to refine, perfect and soon demonstrate the concept of a Terabit Ethernet platform as an aggregation technology for 10 Gig networks in the corporate environment.

These "super networks" may be seen as overkill right now, but with the advances in voice (VOIP) and demands for high availability and ever faster response times, the technology will soon come into its own.

If the industry is incredulous at the Terabit Ethernet data transmission speeds, then these experiments should open their eyes even further: Last year Siemens and WorldCom said they had reached a new speed record of 3.2Tbps for data transmission over a real-world fibre optic network.

Now, NEC claims to have achieved a transmission capacity of 6.4Tbps in the laboratory, making it one of the front-runners in the record-breaking stakes.

But perhaps more industry players should take note of groundbreaking developments by an innovator called Steve Taylor.

President of US-based Distributed Networking Associates, he speculated several years ago on the promise of a new technology called Tachyon Transmission Mode (TTM).

A tachyon is a theoretical particle that can travel faster than the speed of light.

The beauty of TTM - when employed in a theoretical networked environment - is that it operates so fast, you could expect information to be received at the destination almost before it gets transmitted from the source. Now that`s fast.