The NGI Study draft final report discusses the topic of architecture renovation. At the heart of the NGI is architectural evolution that improves upon legacy core protocols of the internet by investigating alternative or auxiliary core infrastructures. This would include projects that are aimed at changing the underlying fabric of the internet and the web itself. Many fundamental issues with resilience and robustness can only be fixed at a systemic level, but the inertia to integrate new solutions into existing Internet is too huge. The matters have been aggravated by the fact that many practical workarounds have been found meanwhile to cater for explosive demand. These workarounds have raised cost, complexity, and even further worsened the known fragility and inflexibility of Internet. The complexity of designing a successful architecture upgrade is easily illustrated by the fact that over half of the life time of the internet has already been spent on the (arguably not very successful) move away from IPv4. Therefore, research and tools to assist in the practical transition or migration to new evolved technologies should also be investigated. Significant effort will have to go into understanding and mitigating the many practical aspects of potential transition from the current internet.
Architecture renovation is widely recognised as critically important for the long term vison of NGI, and can provide structural solutions to problems that can only be partially mitigated within the current architecture. NGI Final report advocates the need for catalysing architectural renovation movement.
Why? Rewriting the ground rules of Internet has a huge effect on higher levels. New ways of Internet addressing that prevent spoofing, a distinct possibility with IPv4, should make it impossible to subsequently brute force attacks. This would for instance change the whole dynamics of currently persistent threats – such as distributed denial of service by botnets. Similar changes can be envisioned to address other key issues such as mass surveillance capabilities. Modern usage is completely different from the static setup of those days, and key features such as disruption tolerance should not be an afterthought. Research into infrastructure renovation is essential to facilitate the introduction of exciting new capabilities apart from mitigating existing limitations. The more fundamental the new characteristics, the further we may evolve the internet.
How? Projects within this banner should investigate fundamental contributions to solving the internet’s challenges. They may question the whole technology stack, the only condition being that they are feasible in terms of technology roll-out and prove their potential to provide lasting answers in line with the NGI Vision. In some cases it will be possible to retrofit novel principles into today’s internet, or to encapsulate current behaviour as an application of the future architecture. In other cases this will be wholly impossible or extremely inefficient. In that case, creating forward compatibility by for instance providing suitable abstractions and mechanisms at the level of end points could be helpful.
Apart from addressing the existential threats from the NGI threat catalogue, potential higher level design goals for alternative infrastructures should be-
- Confidentiality and privacy
- Auditable integrity
- Scope isolation of contingencies
- Redundancy and self-repair
- Disruption tolerance
- Smarter asset distribution
- Better real-time behaviour
- Energy efficiency
Without the movement for Internet architecture renovation, following risks are identified.
- Loss of innovation capabilities and competiveness due to unnecessary technological dependencies.
- Raising the overall cost of future upgrades to the core technologies of the internet (technical debt).
- Loss of diversity, privacy, autonomy and choice for users.
- Giving way to alternative next generation internet technologies incompatible with European values.
- The entire Internet threat catalogue.
Improvements to the system have been proposed and even implemented with various degrees of maturity and success over the course of decades. For the architectural renovation to be effective for NGI, infrastructure innovations should take into account how they should be introduced or retrofitted at internet scale. This means technologies should not just exist in a paper, a technical specification at the IETF, or run on an infrastructure test bed in a lab. Their claims and compatibility should be tested in every possible situation in an automated manner. Investing in maintainability is vital to achieve that means – without a strong global deployment strategy inside operating systems, routers and management software, alternative infrastructures do not stand a chance. In addition to such deployments, providing adequate fallback mechanisms is a priority.