Future Computing for Digital Infrastructures

Making the physical world connected to the Internet is a key step towards digitization of economies which is a priority for many countries as it promises significant improvements in process efficiency and cost savings as well as better livelihood of the citizens. Not only the number of connected devices is growing, but also the data generated by them. For instance, an autonomous test vehicle can typically generate a few TBs of data per day. This data, once processed, can facilitate smarter operations such as active traffic management in cities and more informed decision-making, e.g.,  transportation planning based on human mobility patterns. However, due to the lack of processing resources on the devices, it is estimated that up to 90% of the data collected is never analyzed. Computing infrastructures in the edge to cloud continuum are therefore essential towards realizing smarter systems, from smart cities to smart health. Clearly, with the projected growth in the number of devices and the sheer volume of data generated by them, the role of computing infrastructures will only increase. This situation, unless mitigated by smarter solutions and higher degrees of automation, will only be aggravated. While there is a substantial body of knowledge on cloud computing, emerging use cases, e.g., requiring ultra-low latency as in industrial environments or concerns on privacy and dependability require a fundamentally new approach that is based on interoperable, multi-party and open cloud-edge infrastructure as highlighted by the EU’s Manifesto for the Development of the Next-Generation Cloud Infrastructure & Services Capabilities in 2022. In addition to the need for more-efficient design and operation of these computing infrastructures, growing concerns on the energy footprint of computing systems also raise the urgent need for developing new approaches with a sustainability viewpoint considering both energy efficiency and carbon footprint of the developed solutions.

The perspective of future cloud computing will be shaped by collaboration between future data center and edge infrastructure. The main technical reasons for shifting tasks from the cloud to the edge are latency, bandwidth, locality of data, scalability, accessibility, security, and fault tolerance. This collaboration is not only a technical aspect, but also involves business and stakeholder challenges. The ability to leverage such distributed and heterogeneous compute capabilities critically hinges on the availability of programming models for future computing. Offloading data processing, e.g., requires proper modularization of tasks and frameworks that create the fine-grained parallelism and flexibility that can be leveraged by the upcoming interconnected infrastructure. Existing programming models do not scale well enough to the much more complex and dynamic setups of the future and fail to address the particular reliability/privacy challenges, e.g., compared to the mostly static setup in the cloud.

For our workshop, we identify the following five critical gaps to investigate.

Gap 1: Despite a visible interest in edge and future cloud computing, there is no coherent vision that has gained wide adoption by both academia and industry.

Gap 2: Lack of insight of energy and carbon footprint of applications in the Edge-to-Cloud Compute-Continuum, as well as lack of mechanisms to control these footprints by, e.g., exploiting sources of renewable energy.

Gap 3: Missing programming models for future cloud / edge computing that bridge the heterogeneity of the resources and simplify the task of building scalable and dependable applications across a large number of distributed devices.

Gap 4: Missing ‘Full-stack’ agenda that spans across hardware, programming models, software, algorithms to data, services and socio-technical perspectives.

Gap 5: Lack of alignment between research/technology roadmaps/developments and requirements/awareness of societal/environmental roadmaps/visions. Shortage of resources to support fundamental research in the key enablers and how to evolve them to TRL levels where they can leverage the full societal potential.

Our objectives are highlighted as follows:

- Critically reflect on the state-of-the-art technology

- Reveal challenges and new demands

- Identify key requirements from use cases (technology, economic and societal)  

- Outline essential enablers

- Define a tangible roadmap with concrete collaboration, development and deployment milestones