Nowadays, there is a big debate on redesigning the Internet to meet current and future society needs. In this sce- nario, it is important to consider that actual Internet was designed when computing, communication and storage sub- strate resources were very limited. Technology evolution was much slower than today. In this context, some questions are: how technology evolution will affect design this time; is there any impact of accelerated technological evolution in the choice of candidate technologies; how designs can incorporate technological advances that are increasingly more frequent? This paper contends that: technology evolution is a stronger requirement this time; substrate resources evolution needs to be decoupled from software - and virtualized; and research, development and experimentation needs to be online - integrated on the architecture itself. The paper proposes and discusses two scenarios for online development on Future Internet.
Some impacts of technology evolution on the design of a new internet v0.3
1. Some Impacts of Technology Evolution on the Design of
a New Internet
Antônio Marcos Alberti
Instituto Nacional de Telecomunicações - INATEL
510 João de Camargo, Santa Rita do Sapucaí, Minas Gerais, Brazil
alberti@inatel.br
http://antonioalberti.blogspot.com
2. Outline
• Motivation and Contextualization
• The Role of Technology Evolution on FI Design (FID)
• Two Scenarios to Accelerate Technology Evolution
• Requirements and Challenges Towards the Vision of Online
R&D
• Conclusion
3. Motivation and Contextualization (1/4)
• The Internet was designed at a time where substrate
technologies, such as digital computing, communication and
storage resources, were very limited.
• The first packet switching test of ARPANET was done in a TX-2
computer with 64K 36-bit words of core memory.
TX-2 Module. Source: Wikipedia.
4. Motivation and Contextualization (2/4)
• In 1973, ARPANET had 40 nodes and the assumption was that it
would not have more than 256 networks! [Leiner 2009]
• Soon this estimate proved to be conservative.
• Even in the early 80‘s, the number of hosts on ARPANET already
presented a exponential trend.
Question: Is it possible to predict how
Internet numbers will evolve?
5. Motivation and Contextualization (3/4)
• Moore’s Law - Used by decades to predict technological
developments in terms of computing capacity.
• Ray Kurzweil - Law of Accelerating Returns - two positive
feedback loops occur during some technology evolution process.
Question: Is the current Internet prepared
to face such exponental growths?
• Redesign is required.
6. Motivation and Contextualization (4/4)
• However, this time the technology landscape is quite different
from the 60’s - it is much more dynamic, complex and diverse.
• Objectives:
Question: How technology evolution will affect
design this time?
Question: How designs can incorporate
technological advances that are increasingly more
frequent?
7. The Role of Technology Evolution on FID (1/2)
• Question: How technology evolution will affect design this time?
• Technology evolution will deeply affect designs this time - the pace
of change is impressive.
• To accomodate technology evolution in a flexible, homogeneous,
holistic, and convergent way is a key issue!
• Besides, challenges that can not be addressed today, could be in a
few years - estimatives need to consider exponential growths.
• Impacts of evolution are dual-faceted: they create not only more
capacity, ubiquity, and connectivity, but also more scalability and
complexity drawnbacks.
8. The Role of Technology Evolution on FID (2/2)
• Question: How designs can incorporate technological advances that
are increasingly more frequent?
• Making R&D cycles online - the vision of Online R&D:
1. Indirect R&D over a developing facility compatible with an operational
substrate resource provider.
2. Direct R&D over an operational substrate resource provider.
3. Indirect R&D over a third party experimental facility compatible with an
operational substrate resource provider.
11. Requirements and Challenges Behind Online R&D
• All hardware needs to be virtualized - however, FI research is not so
generic.
• The partners of online R&D need to trust each other - a trustable
R&D business ecosystem needs to be structured.
• Security and privacy need to be rethinked on the context of such
ecosystem.
• Real-time resource virtualization is important to enable deployment of
time-sensitive functionalities on software.
• Current autonomic initiatives are little concerned with accomodating
acclerating technology evolution - FI requires better evolvability and
sustainability.
12. Conclusion
• Online R&D accomodates technology evolution of substrate
resources, software systems, virtual entities, and frameworks.
• The main advantages are flexibility, transparency, conformity to the
user requirements and proximity to the operational scenarios.
• Virtual operators can participate at distance from milestones on R&D
schedules.
• R&D of new substrate resources consider virtualization of their
functionalities to operational parters in the business ecosystem.
13. Conclusion
• Online R&D of customized networks can be done over slices of
virtualized resources.
• When R&D ends, the customized network is simply made operational.
• The online R&D facilities evaluation, comparison, migration and
evolution of substrate and overlaid resources.
• In summary, designers and developers of the FI are also active part of
the digital business ecosystems they are creating.