Energy efficient data centres
Over the last 5 decades, a wide range of different computing platforms have been developed, establishing themselves for several applications. Most require specific data centre architectures to optimize their operation for improved resource efficiency or comprehensive environmental sustainability. Research and innovation on data centers intend to combine specific computer systems and their incumbent infrastructure (such as cooling systems and connection to the power grid) with novel algorithms in ways that can solve the many challenging problems. Innovation potential lies in the computer systems, their architecturesand infrastructure, the algorithms, and requirements (namely, efficiency, sustainability, security, sovereignty, robustness, and more).
12th International Workshop on The ever-evolving data centers: research and innovation (E2DC 2024)
There are two themes that deserve special attention within this workshop. First, the phenomenal growth of AI applications has pushed HPC computing towards the mainstream. The increased challenges for the data centre infrastructure are leading to a broader interest in advanced architectures and technologies such as liquid cooling techniques. Second, is the growing needs for innovative techniques and solutions for data centres located in environmentally challenging locations.
Important Dates
March 31st , 2024
April 19th , 2024
Paper submission
April 26th , 2024
Notification of authors
May 3rd , 2024
Camera-ready papers
Call for Papers
The workshop invites original papers up to 8 pages that were not published before and are currently not under consideration and/or review for publication elsewhere. Topics include, but are not limited to:
- Novel or enhanced data center architectures and infrastructure
- Specialized by application driven requirements.
- Specialized by environmental conditions and constraints.
- Improved resource and energy efficiency
- Resilience against unforeseen environmental impacts
- Large-scale and small-scale architectures
- Hybrid data centers
- System perspective of data centers in digital infrastructures or federated service ecosystems
- Operating and optimizing heterogeneous data centers
- High Performance Computing and Network Computing
- Cross-site cooperation in data center networks
- Combination of Edge Computing and Advanced Computing
- Different computation variations and influences on DC architectures
- Different DC architecture combinations and improved service performance
- Improvements on efficient data processing
- Improvements on real-time data processing
- Implications on systems and service engineering
-
Innovations within or by data centers
- Recombination of established technologies
- New data center designs for new applications
- Improved sector coupling
- The role of data centers within innovations
Submission and publication of the work
The workshop will be co-located with ACM e-Energy 2024 Please submit your manuscript via submission portal. Manuscripts will be published in the ACM proceedings and must adhere to ACM format. Submission implies the willingness of at least one author to register and present the paper at E2DC. The organizers reserve the right to exclude a paper from publication if it is not presented. For more information please contact the organizers: Ariel Oleksiak (ariel at man.poznan.pl), Gunnar Schomaker (schomaker at sicp.de)
KEYNOTE SPEAKER
Poh Seng Lee,
National University of Singapore
The keynote presents an exploration of the advancements and strategies reshaping the future of data centers. It will delve into how the integration of innovative cooling technologies, such as direct-to-chip and immersion cooling, alongside sustainable architectural designs, are critical in meeting the increasing demands imposed by AI technologies and extensive data processing. Highlighting real-world applications and case studies, the talk will illustrate the significant efficiency gains and environmental benefits these technologies offer. The presentation will also address the broader implications of these technological advancements, including the necessary transformation of workforce skills and the integration of AI and IoT into data center management. The keynote provides a vision on how data centers can evolve to be more resilient, efficient, and aligned with environmental sustainability goals in the AI era.
E2DC WORKSHOP PROGRAM 2024
9:30 am - 10:45 am
Welcome and Opening remarks
Ariel Oleksiak (Poznan Supercomputing and Networking Center)
Keynote:
Fostering Resilience and Innovation in Data Centers: Embracing Next-Generation Cooling and Sustainable Architectures for the AI Era
Poh Seng Lee (National University of Singapore)
Jonas Gustafsson, Jon Summers (RISE), Joaquim Romaní, Santiago Escudero (IREC)
10:45 am - 11:15 am
Break
11:15 am – 12:30 pm
E2DC-NTU Special Session on AIDC
AI and Digital Infrastructure for a Sustainable Future: A Tale of Twin TransitionYonggang Wen (College of Computing and Data Science, NTU)
Direct Liquid Immersion Cooling in Thermal Management of Data Center
Fei Dan (School of Mechanical and Aerospace Engineerin, NTU)
Design of Efficient and Transparent Schedulers for Deep Learning
Tianwei Zhang (College of Computing and Data Science, NTU)
12:30 pm – 2:00 pm
Lunch
2:00 pm - 3:15 pm
Industrial keynote & discussion: The end in mind – taking stock of how we build and operate data centres
Data Center CO2 emissions are still not a currency! The desire for a uniform methodical implementation of the life cycle assessment of waste heat utilization in data centers.Gunnar Schomaker (Universität Paderborn)
Models and simulation software for analysis and optimisation of data centers waste heat re-use potential
Ariel Oleksiak (Poznan Supercomputing and Networking Center)
Metrics for Data Center Sustainability
Jon Summers (RISE)
3:15 pm – 3:45 pm
Break
3:45 pm - 5:00 pm
Automation of data center airflow and heat transfer analysis for transient scenarioW. Szeliga, A. Oleksiak (PSNC) R. Roszak (PUT)
An Optimization Approach for Selection and Deployment of Energy-Efficient Servers in Green Data Centers
Duc Van Le and Rui Tan (NTU)
Hybrid Heterogeneous Clusters Can Lower the Energy Consumption of LLM Inference Workloads
Grant Wilkins, Richard Mortier (University of Cambridge), Srinivasan Keshav (University of Cambridge, University of Waterloo)
Workshop Chairs
Ariel Oleksiak
Poznan Supercomputing and Networking Center, Poland
ariel@man.poznan.pl
Gunnar Schomaker
Software Innovation Campus Paderborn, Germany
schomaker@sicp.de
Jon Summers
RISE Research Institutes of Sweden, Sweden
jon.summers@ri.se
Technical Programme Committee
Poh Seng Lee
National University of Singapore
Robert Basmadjian
TU Clausthal, Germany
Riccardo Pinciroli
Gran Sasso Institute, Italy
George Da Costa
University of Toulouse, France
Xavier Hesselbach-Serra
Universitat Politecnica de Catalunya, Spain
Jonas Gustafsson
Research Institutes of Sweden, Sweden
Tudor Cioara
Technical University Cluj-Napoca, Romania
Boudewijn R. Haverkort
Tilburg School of Humanities & Digital Sciences
Hongyang Sun
University of Kansas, United States of America
Suresh Subramaniam
The George Washington University, United States of America
Guru Venkataramani
The George Washington University, United States of America
Evgenia Smirni
College of William and Mary, United States of America
Hermann de Meer
University of Passau, Germany
Natalie Bates
Energy Efficient High Performance Computing Working Group, United States of America
Contact
Ariel Oleksiak (ariel at man.poznan.pl)
Gunnar Schomaker (schomaker at sicp.de)
Jon Summers (jon.summers at ri.se)