We focus here on the approach that the project is taking for going after the expressed challenges. For more detail on the actual work, e.g., design choices, implementation, etc, please refer to the public deliverables or the information in the dissemination and software section of this website.
PURSUIT, similar to its predecessor PSIRP, is an architecture design project that attempts to build solutions for a new form of internetworking. The project intends to undergo all the stages from early evaluation of existing work to design of a solution that is intended to be very different from what we know today in our networks. Within the PSIRP project, we established and successfully tested a design methodology that has proven to be effective in a project of the size of PSIRP (and PURSUIT) as well as accommodate the desire of individual researchers to perform world-class research. Here, we outline this approach that has been adopted for PURSUIT, too.
Combining Bottom-Up and Top-Down
The major challenge in PURSUIT is to create the right combination of bottom-up ideas and top-down rationalization (which can be seen as forming a somewhat rigorous design process). We do recognize that the bottom-up approach is of utmost importance to push the limits within the goals of our project. Different ideas are taken, tested, integrated but also discarded. The team structure of PURSUIT encourages this bottom-up approach and it is expected that in particular design choices are pushed forward in this manner.
However, we also recognize the importance of rationalizing our work in an attempt to clearly formulate our ideas against a rationalizing and well-founded design process. This is seen as helpful to carve out the essence in our propositions, steer towards areas of necessary investigation, and also document the rationale with which certain decisions and choices can be considered to have been made with.
This view on our approach, as a combination of bottom-up work and top-down rationalization, leads to the following methodology for our architecture design.
Our intention of following a clean slate design approach can be recognized by the formulation of a clear vision (one which is larger than the scope of the project) and the definition of goals for our design. We then follow suit by clearly outlining the principles for our design from which we derive design considerations and patterns to be seen throughout the overall system. This will lead us to a conceptual architecture and its components, to be mapped onto (several) design choices to be implemented and evaluated against the particular requirements of the choices, the requirements being derived from the governing principles of our design.
While this approach seems rather top-down and sequential at first sight, one can recognize the intention to permanently adapt the design process by virtue of allowing any step of the design to be questioned, leading to refinements at any level. In particular issues like scalability - often hard to grasp as top-down design constraints – are seen to be driven by this bottom-up experience when it comes to the particular fields that require attention in this regard. This questioning indicates the integration of the bottom-up approach that will guide our daily work on the particular choices we make.
It is worthwhile noting that one might not recognize a typical top-down requirements engineering process in this approach. This is intentional since strict requirements at the very beginning of our design are seen as being too restrictive. Instead, we base our design on a visionary approach with design principles that will allow for different design choices to emerge, all implementing our vision and principles but potentially all following very different detailed approaches in achieving this. This is complemented by our lessons learned from the implementations (the combination of top-down and bottom-up), leading to a more evolutionary and holistic approach where specifications of components are integrated rather late on component level while the overall systemic view is preserved, guided by rough design principles as well as questioned and therefore evolved by our learning and progress.
Life Cycle within our Methodology
While the above methodology describes micro cycles of development throughout the design process for an architectural solution, PURSUIT also implements a macro cycle. For that, we follow an iterative approach from brainstorming (around principles, design considerations and choices) over the identification of structures and patterns to rapid prototyping and evaluation of ideas, leading to a refinement of the original design. Hence, this process implements a single macro life cycle (and feedback loop) for our design methodology above and will effectively lead us to a collection of refinements and choices throughout the project lifetime. This macro life cycle has been the basis for our project planning and is therefore the foundation of our project structure, reflected in the team structures and the timing of the deliverables.
Is this ‘Clean Slate’ or Not?
Often, the term clean slate is used to describe what PURSUIT intends to do. We do not intend to focus on whether or not PURSUIT is indeed clean slate. The focus of our view on clean slate is that of undergoing a rigorous design process, leading us towards a set of possible solutions for laying the ground for the Future Internet. That view steers us towards the direction of formulating this design process before continuing in particular directions of design choices.
However, it is important to stress that PURSUIT takes indeed a stand on questioning ALL fundamentals of today’s Internet, re-investigates them in the light of an information-centric view, and design solutions that might suit better our re-evaluation. In that sense, PURSUIT is indeed assuming a clean slate (executed within the above formulated design methodology). However, we do recognise the necessity to embed any solution architecture into an evolvable story from here to there. Hence, socio-economic factors, migration and deployment strategies as well as legacy compliance are crucial work items in our work. We call this clean slate with grounding to reality.