Communications Insider: June 2009
IPv6: An Historical Perspective (Hindsight is a wonderful thing)
John Thomson
Director, Software Engineering
GE Fanuc Intelligent Platforms
It may well be that the story of Bill Gates claiming that "640K is enough memory for anyone" is apocryphal. IBM's Thomas Watson may not have said in 1943, as was widely claimed, that he estimated the total worldwide market for computers to be no more than five. Ken Olson however, the founder of DEC, certainly offered a hostage to fortune when he told a convention of the World Future Society in 1977 that "there is no reason for any individual to have a computer in his home" – even if his remark has been taken out of context. The point is that hindsight – especially in the world of technology – is a very wonderful thing, not to mention a very precise science. History has a way of making us look stupid.
Take communications, for example – an area of technology that has made unbelievable strides since Gates and Olson are supposed to have made their pronouncements. At a time when we take instantaneous access to any computer in the world for granted – and not just from our desks or homes, but wherever we are and whatever we're doing – it is easy to believe that it has always been this way. But in fact, the path to where we are today has been one characterized (with the benefit of hindsight) with missteps, with journeys along blind alleys and with an almost complete, if entirely understandable, inability to see where the world might be going.
How, though, did we get to where we are today? Back in the early eighties, when the PC was only just emerging from a plethora of dedicated word processors, and 'small business computers' (which were about the size of a desk, and had the processing power of a modern-day wristwatch) the electronic communications scene was relatively limited. There was the telephone network – huge, complex, and worldwide, with a few international standards, pulling together systems that were largely state-run. These standards were handled by the CCITT (Comité Consultatif International Téléphonique et Télégraphique), a UN body which later morphed into the ITU (International Telecommunication Union) In addition to the telephone system, there were a few other general means of electronically communicating with others: most people knew of telegraph and telex services, but that was about the extent of general knowledge. Dedicated 'bare wire' type circuits or modem-equipped data circuits existed, but were uncommon, very specialist, and generally restricted to organizations like banks. These were mainly used for mainframe computer systems to talk to their dedicated terminals.
New ideas, however, were starting to come on to the scene. In-house systems – long the preserve of basic accounting functions such as ledgers, stock control, invoicing and payroll - were being developed for applications such as email (although email as a term was not in widespread use, with proprietary alternative names – such as the Texas Instruments 'MSG' system – more common). Ways of connecting together multiple mainframes were developing, and many of the computer manufacturers were starting to develop their various 'network architectures'. The dominant example of this was IBM's SNA (Systems Network Architecture) – an example which other mainframe manufacturers, such as Burroughs with its BNA, were quick to follow.
Within buildings, the wiring schemes used were many and various – using a mixture of technology types. The idea of a Local Area Network (LAN) was the hot topic on many people's lips, but much development effort was focused on the growing requirement to connect intelligent (or, at least, somewhat intelligent) terminals to a mainframe. In a parallel effort, the academic community was the principal proponent of something called the Internet Protocol (IP) – but this was of only slight interest to the majority of the computer industry.
So, we could see the beginnings of three groups – the wide area group (such as CCITT), the local area group (typically computer manufacturers) and the IP group (who were mainly academics). These three groups would go on to dominate (and at times squabble over) the development of communications standards for the next three decades. Hindsight has left the IP group – with its focus on a smaller set of requirements and its pragmatic approach to problems - in the most favorable light. History has been less kind, on the other hand, to SNA: it was an innovative and very sophisticated architecture and technology in its time but, like so many proprietary computing standards, it has failed to endure. IP is now ubiquitous: SNA is almost nowhere.
Eventually, the industry – individually and collectively; the mainframe/LAN people, the telecommunications/WAN people and the academic community – realized that for real progress to be made, it was vital to establish some common ground for future development. This realization led to the creation of the now-famous OSI (Open Systems Interconnection) reference model. Through many years of seriously hard work, a number of the world's experts worked on coming up with a way of explaining how signals on a length of wire could represent meaningful information across frighteningly complex networks. This resulted in two outcomes. The first was a useful reference model (the 'seven layer' model) which is now the general way used by both the computing and the telecommunications worlds to talk about networks. The second was a set of protocols, designed to reflect the model's structures. Most of these protocols are gone and forgotten, having been replaced by their IP-based equivalents.
The OSI protocol suite had many strengths and weaknesses – which are still the subject of debate today. One of those debates at the time was one of the key areas that the protocol suite covered: the topic of addressing.
Looked at from the point of view of the telecommunications/WAN group, the addressing scheme proposed by the academics – the Internet Protocol – seemed wholly unworkable. The idea that only 32 bits were available for addressing seemed incredibly limiting. In fact, because of the subnet structuring imposed on it, the range was effectively far smaller than a simple 32-bit number. These were two radically divergent communities in terms of their world view. While the telecommunications people were thinking of possible end points as some multiple of the world's total population, thinking in the IP world was dominated, implicitly if not explicitly, by a view of the potential size of the computing market that saw Thomas Watson's and Ken Olson's supposed remarks become the stuff of legend.
But 32-bit addressing, with its address space of just over four billion unique addresses, was accepted and agreed, and IPv4 – the fourth iteration of the Internet Protocol, and the first to be widely deployed – became the basis on which the networks of the future would be built. If every man, woman and child in the developed world in 2008 had access to just a single telephone, and the world's population had remained static at those levels, four billion addresses might just have been enough.
But since IPv4 became official in 1981, there has been an explosion in the use of networks. Had networks been limited to just telephony, it's unlikely that four billion addresses would have been enough – without taking into account the proliferation of networkable computing devices, both static and mobile. The shortage of IPv4 addresses is becoming – has become – a significant problem. There have been pragmatic workarounds – primarily in the area of NAT (Network Address Translation) mechanisms - and it is these that are currently underpinning the Internet today.
It is now easy to look back and realize that the adoption of 32-bit addressing represented very shortsighted thinking. Beyond this, the workarounds in place today, while effective, are no basis on which to build the networks of the future. All hail, then, IPv6 with its 128-bit addressing - with a range of powerful, flexible allocation schemes. Of course, IPv6 is far more than just a longer addressing scheme, and most of the other additions are also new requirements. Some of the security threats, for example, that are now commonplace were not even science fiction back when IPv4 was being developed. IPv6 therefore incorporates facilities for security such as encryption, authentication and so on. IPv6 has also added features of interest to the mobile community – a community that was almost unimagined thirty years ago.
Although it is estimated to have a penetration of less than 1% worldwide today, IPv6 is the basis on which next generation networks are being designed. In fact, it is a requirement for a large part of the military market, and is a growing requirement in much of the telecoms market. That's why GE Fanuc offers a broad range of advanced IPv6-capable NETernity network switches.
Increasingly, network designers and users need to have migration plans in place, and for many, IPv6 is now a daily reality. The co-existence of IPv4 and IPv6 networks, and the ability to interwork between them, is becoming the norm. Products that ease that interworking and migration are in great demand. Techniques for 'tunnelling' IPv6 traffic over an old IPv4 network allow new, IPv6 systems to talk to each other, while sitting on current networks. It is unlikely to be long before the situation is reversed, and it will be IPv4 systems that will be in the minority, sitting on top of IPv6 infrastructure. Making this seamless is one of the dreams of network designers, and is being facilitated by network components like 'dual-stack' systems – where both IPv4 and IPv6 are run on the same host. Switches and routers are providing various flavors of tunnels, with terms like "6to4" and "6over4" starting to have real meaning.
There is much work to do yet, before the last ever IPv4 packet gets transmitted, and it has only been recently that IPv6 addresses have been served by the public Domain Name Servers. The dual-stack world and NAT will be with us for some time to come – but IPv6 will enable the current explosive growth in networks and network-capable devices to continue unabated. Or, with hindsight, will that statement rank alongside those of Gates, Watson and Olson?
|
