It’s been said (and often repeated) that those who don’t learn the lessons of history are doomed to repeat them. I believe this also holds true for Internet history, especially as we make the transformation to the hyper-connected digital society.
A 30th anniversary article in the Globe and Mail’s Report on Business (ROB, March 2015) described the pioneering efforts to establish the U.S. government’s ARPANET and then the Internet. Even though the basic concepts were developed in the 1960’s and 70’s, the first domain name was not registered until March, 1985. Today, with over 100 million domains, the Internet has become one of the big success stories of the twentieth century. Vint Cerf, a co-inventor of the original network protocols, is now cast as the Alexander Graham Bell of the modern era.
For those interested in the history of networking from the 1970s to 1990s, there are some very useful references. For example, check out:
- Open Standards and the Digital Age by Andrew L. Russell (Cambridge University Press, 2014)
- Patterns in Network Architecture – A Return to Fundamentals by John Day (Prentice Hall, 2007)
- INWG and the Conception of the Internet: An Eyewitness Account by Alex McKenzie
There should be useful lessons from the Internet’s development, both technical and commercial, that can be applied to cloud computing and the Internet of Things. Here are five suggestions:
- Consensus standards are necessary but not sufficient
- Common principles and practices should be clearly defined
- User demand and funding create a sustainable ecosystem
- Security, privacy, location awareness and identification must be foundational
- Innovation and change must be managed
Standards are required
Since the beginning, the Internet has been based on the mutual agreements of a set of researchers and engineers. The Internet Engineering Task Force (the IETF) continues to guide the formalization and adoption of Internet documents. The real lessons are that accepted standards are essential, but that standardization should not be used mainly to create competitive advantage or support product marketing for any single stakeholder.
One section of the ROB article caught my attention:
“…This adopted smoothness [referring to his 3 piece suits] proved helpful when a competitor to TCP/IP, the so-called Open Systems Interconnection, emerged. OSI was popular in Europe – produced, the story goes, in a more orderly fashion by scientist’s working at the behest of bureaucrats. To many, the protocol felt more secure.
As DARPA’s representative, Cerf has the unpleasant duty of attending OSI conferences, biting his tongue, cordially disagreeing that the upstart was all that…..”
This refers to the Reference Model for the Open Systems Interconnection (OSI) and the various protocols that were being developed by the International Organization for Standardization at roughly the same time as TCP/IP. At that time, there was significant competition among standards developers. Ultimately, there was little benefit derived from the conflict.
As the Internet expands its reach and becomes more software-defined, the basic questions will be: Will new standards be required, and if so, what should they include, who should develop them, and how will a comprehensive architecture be defined?
Common principles and practices
The Internet of the future will have a significant impact on individuals, on economies and even on societies. It will need to be very widely understood, universally accepted, technically sophisticated, and highly responsive to user requirements.
Common principles and practices for Internet design, operation, protection, and management are essential starting points for the anticipated Internet transformation. The history lesson is that these are difficult to achieve a priori without co-ordinated planning and recognized governance authorities.
Some of the better-known Internet principles are:
- End-to-end principle: application-specific functions should be implemented at the network endpoints rather than in intermediate nodes (i.e., simple network, complex hosts);
- Best efforts/net neutrality: the general principle of “best efforts” delivery is inherent in datagram-based networking; and
- Rough consensus and running code: the practice of requiring majority agreement and at least two versions implemented for a standard to be adopted.
There are various other principles and many underlying practices that need to either be preserved or explicitly be replaced in the future Internet. The lesson for the multi-user, multi-purpose, multi-national Internet is that the basic assumptions should not be implicit or restricted to an exclusive group of experts.
User demand and funding
The Internet did not spontaneously appear or grow organically. University research projects, largely funded by the U.S. Department of Defense via its (Defense) Advanced Research Projects Agency (ARPA), developed and tested the concepts and technologies that led to the Internet as we know it today.
Everything from early packet switching in the 1960s to the TCP/IP protocols in the early 1980s can be traced to ARPA sponsorship. See the references listed above and the Internet Society history page.
In 1980, as further indication of funding, we have:
“(D)ARPA needed a team to implement its brand-new TCP/IP protocol stack on the VAX under Unix……DARPA considered contracting DEC to implement TCP/IP but rejected that idea…..Instead, DARPA chose Berkeley Unix as a platform — explicitly because its source code was available and unencumbered.”
The TCP/IP implementation for Unix was released with Berkeley 4.2 in 1983, which meant that the TCP/IP protocols were freely available for all to use.
The lesson to be learned is that the Internet resulted from the demands of a very significant user (the U.S. Department of Defense) with sufficient funding to create demonstrable solutions (again, from DoD via ARPA) and the mandate to support their use in universities.
It is not clear that equivalent conditions and support exists for the development of either cloud computing or the Internet of Things. The major issues associated with the migration to IPv6 provide a good example and analysis of how the world of networks has changed! A Cisco Journal article also discusses issues with the Internet.
Security, privacy, location and identification
It isn’t feasible to cover all the potential (and required) lessons that the Internet could provide in the areas of security, privacy, location and identification. Suffice it to say that many people believe security was an afterthought when the Internet was first developed.
Perhaps the lesson to be learned is that these four areas must be included in the foundation of any next generation digital environment. We need only look at today’s security breaches, email spam, identity thefts and network attacks to recognize the importance of knowing who is using the network, what they are doing, where they are, and what they are accessing.
The initial design of the Internet would have been less concerned with security and privacy due to the limited initial applications and the implicit trust that existed among the users. The future hyper-connected Internet, with potentially more than fifty billion attached “things” is a very different scenario.
The 1980s Internet generally had relatively fixed access points and communications was point-to-point. Today, everything – people, cars, trains, etc. – is constantly moving and changing, and there is a much greater need for users to be identified, tracked, and verified. Managing and reporting on the reputation of the connected “things” has become important to ensure that data is correct and actions are allowed.
Innovation and change management
One final lesson from the Internet – build technologies and solutions with obsolescence and extensibility in mind. As we are well aware, both the user requirements and the provider technologies will change over time. Who would have guessed we would need to support live video (such as the Meerkat app) or internet-enabled appliances such as stoves and washers?
Let’s hope that cloud computing and the Internet of Things/Everything are not the stick that breaks the elephant’s back as far as the Internet is concerned!
