Periodic Reporting for period 1 - CLOUDSTARS (Cloud Open Source Research Mobility Network)
Reporting period: 2023-01-01 to 2024-12-31
CLOUDSTARS presents a joint research programme in the fields of Cloud computing and AI technologies. On one hand, CLOUDSTARS pursues innovation in the Cloud infrastructures to support the next generation of low-latency, high-performance complex workloads which are now being developed. On the other hand, the research programme pursues high innovationin the Cloud, making the best application of artificial intelligence techniques and AI models with automatic adaptation to the computing resources both in the Cloud and the Edge to optimize where data are being processed.
The participants in CLOUDSTARS exchange skills and knowledge to allow them to progress towards key advances in cloud computing, data analytics and AI technologies, while strengthening collaborative research between different countries and sectors. In particular, the staff members who participate in the project develop new skills (open-source contributions, access to cutting-edge Cloud/Edge technologies), be exposed to new research environments.
To facilitate the exploitation of results, CLOUDSTARS will favor industrial open-source technologies whenever possible. This will increase the overall global impact of research contributions, helping to arrive to millions of interested third parties through open-source communities. In this sense, the companies involved in the project will transfer knowledge and open-source projects to their commercial products and cloud offerings, providing a great economic and social impact.
The participants in CLOUDSTARS exchange skills and knowledge to allow them to progress towards key advances in cloud computing, data analytics and AI technologies, while strengthening collaborative research between different countries and sectors. In particular, the staff members who participate in the project develop new skills (open-source contributions, access to cutting-edge Cloud/Edge technologies), be exposed to new research environments.
To facilitate the exploitation of results, CLOUDSTARS will favor industrial open-source technologies whenever possible. This will increase the overall global impact of research contributions, helping to arrive to millions of interested third parties through open-source communities. In this sense, the companies involved in the project will transfer knowledge and open-source projects to their commercial products and cloud offerings, providing a great economic and social impact.
CLOUDSTARS focuses on research in three scientific fields divided into work packages. The activities performed are classified according to the corresponding field.
- Next-Gen Cloud Infrastructure (WP2): Development and benchmarking of open-source technologies to advance the next generation cloud infrastructure.
- Serverless and Hybrid Cloud Application Design (WP3): Investigation of mechanisms to improve scalability and elasticity of hybrid cloud applications along with performance optimization. Understanding the potential characteristics of workflow steps for optimizing technologies such as Kubernetes and novel 'databaseless' Calcite and contributions to the open source GEDS project for optimizing the functionality of ephemeral data store events.
- AI and Data Analytics (WP4): Evaluation of platforms and systems for hosting Large-Language Model (LLM) training and inference provides insights into the performance of different strategies used to optimize data ingestion in AI applications. Development of an efficient serverless-based framework for Big Data and AI processing from an automatic selection of planners created with Graph Neural Networks and research on cluster scheduling for deep learning workloads.
- Next-Gen Cloud Infrastructure (WP2): Development and benchmarking of open-source technologies to advance the next generation cloud infrastructure.
- Serverless and Hybrid Cloud Application Design (WP3): Investigation of mechanisms to improve scalability and elasticity of hybrid cloud applications along with performance optimization. Understanding the potential characteristics of workflow steps for optimizing technologies such as Kubernetes and novel 'databaseless' Calcite and contributions to the open source GEDS project for optimizing the functionality of ephemeral data store events.
- AI and Data Analytics (WP4): Evaluation of platforms and systems for hosting Large-Language Model (LLM) training and inference provides insights into the performance of different strategies used to optimize data ingestion in AI applications. Development of an efficient serverless-based framework for Big Data and AI processing from an automatic selection of planners created with Graph Neural Networks and research on cluster scheduling for deep learning workloads.
The major outcomes of the project become part of global widely adopted open source Cloud technologies. Research produced in secondments is incorporated to major open source projects in the field and promoted by the five participating industrial partners and the academic network, which produces direct economic impacts.
The project greatly contributes to the following key scientific impacts:
1. Impact on Cloud Infrastructure Management and Optimization
2. Impacts on Serverless and Hybrid Cloud Application Design
3. Impacts on AI and Data Analytics
There are world-wide scientific impacts in the three aforementioned fields implying top research publications in scientific journals and conferences. The potential of joining top academic talent in distributed systems with state of the art Cloud and Edge Infrastructure certainly produces top and highly cited research publications in the field.
The project produces very relevant societal impacts in European societies. The transformation of European societies thanks to digitisation produces relevant impacts in different fields. Cloud Computing, Edge Computing, and Artificial Intelligence are the backbone technologies required for the digitisation of the different sectors. Since CLOUDSTARS partners already participate in different EU research projects and national initiatives, the consortium will be able to demonstrate societal impacts in a variety of domains and use cases like Health, Agriculture, Industry, or Transport to name a few.
The project greatly contributes to the following key scientific impacts:
1. Impact on Cloud Infrastructure Management and Optimization
2. Impacts on Serverless and Hybrid Cloud Application Design
3. Impacts on AI and Data Analytics
There are world-wide scientific impacts in the three aforementioned fields implying top research publications in scientific journals and conferences. The potential of joining top academic talent in distributed systems with state of the art Cloud and Edge Infrastructure certainly produces top and highly cited research publications in the field.
The project produces very relevant societal impacts in European societies. The transformation of European societies thanks to digitisation produces relevant impacts in different fields. Cloud Computing, Edge Computing, and Artificial Intelligence are the backbone technologies required for the digitisation of the different sectors. Since CLOUDSTARS partners already participate in different EU research projects and national initiatives, the consortium will be able to demonstrate societal impacts in a variety of domains and use cases like Health, Agriculture, Industry, or Transport to name a few.