Periodic Reporting for period 1 - ECO-Qube (Artificial-Intelligence-Augmented Cooling System for Small Data Centres)
Reporting period: 2020-10-01 to 2022-03-31
Information and communication technologies are experiencing a paradigm shift called "edge computing". Emerging digital trends – from artificial intelligence (AI), augmented reality (AR), and virtual reality (VR) to the Internet of Things (IoT) – are making data centres more latency and security sensitive while forcing data centres to be located closer to end users. However, current technologies for hyper scale data centres are not efficient for edge computing systems. Consequently, edge data centres are challenged to keep up with new energy efficiency measures.
Quantitative performance and efficiency measurements and risk mitigation strategies are the weakest points for edge computing systems. In other words, most of the edge computing end users do not know how they can enhance the performance and efficiency while reducing the energy consumption and carbon footprint of their data centre. Moreover, most of the existing hardware and software technologies work independently and reducing the overall efficiency of the systems.
Artificial-Intelligence-Augmented Cooling System for Small Data Centres “ECO-Qube”; is a holistic management system which aims to enhance energy efficiency and cooling performance by orchestrating both hardware and software components in edge computing applications.
ECO-Qube is a data driven approach which utilizes valuable unused data from active data centre components. Created big data is being used by an artificial intelligence augmented system which detects cooling and energy requirements instantaneously.
ECO-Qube differentiates from conventional cooling systems which keep operating temperatures within a strict interval and do not evaluate measurable cooling performance. Unmeasured cooling performance leads underperformed airflow, thermal disequilibrium, and high energy consumption. On the contrary, ECO-Qube offers a zonal heat management system which benefits from Computational Fluid Dynamics (CFD) simulations to adapt cooling system for the best airflow and cooling performance with minimum energy consumption. Moreover, ECO-Qube realizes smart workload orchestration to keep the CPUs at their most energy efficient state and maintain the thermal equilibrium to reduce overheating risk.
Sustainability is another priority for ECO-Qube’s smart energy management system (EMS), which is designed to track the energy demand and operate the energy supply in cooperation with building/district’s EMS. This synergy maximizes the energy supplied from renewable energy sources and minimizes the energy supplied from sources with big carbon footprint.
Quantitative performance and efficiency measurements and risk mitigation strategies are the weakest points for edge computing systems. In other words, most of the edge computing end users do not know how they can enhance the performance and efficiency while reducing the energy consumption and carbon footprint of their data centre. Moreover, most of the existing hardware and software technologies work independently and reducing the overall efficiency of the systems.
Artificial-Intelligence-Augmented Cooling System for Small Data Centres “ECO-Qube”; is a holistic management system which aims to enhance energy efficiency and cooling performance by orchestrating both hardware and software components in edge computing applications.
ECO-Qube is a data driven approach which utilizes valuable unused data from active data centre components. Created big data is being used by an artificial intelligence augmented system which detects cooling and energy requirements instantaneously.
ECO-Qube differentiates from conventional cooling systems which keep operating temperatures within a strict interval and do not evaluate measurable cooling performance. Unmeasured cooling performance leads underperformed airflow, thermal disequilibrium, and high energy consumption. On the contrary, ECO-Qube offers a zonal heat management system which benefits from Computational Fluid Dynamics (CFD) simulations to adapt cooling system for the best airflow and cooling performance with minimum energy consumption. Moreover, ECO-Qube realizes smart workload orchestration to keep the CPUs at their most energy efficient state and maintain the thermal equilibrium to reduce overheating risk.
Sustainability is another priority for ECO-Qube’s smart energy management system (EMS), which is designed to track the energy demand and operate the energy supply in cooperation with building/district’s EMS. This synergy maximizes the energy supplied from renewable energy sources and minimizes the energy supplied from sources with big carbon footprint.
An open source computational model was developed for the simulations of flow and thermal structures in data centers and validated with the experimental data in the literature. The validated numerical model was used for the investigation of thermal and cooling efficiencies of the pilot sites. A new server model was developed for Open Compute Project (OCP) servers in the micro-data center for the prediction of the flow rate of the cold air passing through the servers in a realistic way. A virtual wind tunnel model was developed for the calculation of pressure loss while turbulent flow passing through the server.
ECO-Qube EMS has been developed within this task which details have been presented in D3.1. ECO-Qube EMS will monitor and control both produced and consumed energy within the data centre ecosystem with a holistic approach with ECO-Qube EMS which is developed following SAREF standards.
The System Architecture of the Workload Orchestration System as outlined in the proposal has been defined and iterated on with the involved project partners from other WPs.
WP8 has disseminated ECO-Qube results and promote the project to interested stakeholders which will be the main users of the project results as well as to provide support to set up a network structure for achieving its future sustainability.
ECO-Qube EMS has been developed within this task which details have been presented in D3.1. ECO-Qube EMS will monitor and control both produced and consumed energy within the data centre ecosystem with a holistic approach with ECO-Qube EMS which is developed following SAREF standards.
The System Architecture of the Workload Orchestration System as outlined in the proposal has been defined and iterated on with the involved project partners from other WPs.
WP8 has disseminated ECO-Qube results and promote the project to interested stakeholders which will be the main users of the project results as well as to provide support to set up a network structure for achieving its future sustainability.
ECO-Qube targets 20% PUE improvements versus the same configuration without ECO-Qube. ECO-Qube solution is expected to achieve 20% of PUE and CO2 savings improvement by 20% PES savings and 80% REF and ERF realization with the theoretical maximum values.
ECO-Qube is still in progress and potential impacts of the project has not been so clear yet.
D5.1 ECO-Qube Guidelines for efficient implementation of data centers in buildings will raise wareness among the market and support edge computing market to reach net zero emissions by 2030.
ECO-Qube is still in progress and potential impacts of the project has not been so clear yet.
D5.1 ECO-Qube Guidelines for efficient implementation of data centers in buildings will raise wareness among the market and support edge computing market to reach net zero emissions by 2030.