|
pdsw 2023:8th International Parallel Data Systems Workshop
|
| Program Co-Chairs: Microsoft, USA The Ohio State University, USA |
General Chair: Hong Kong Baptist University, China |
About the PDSW Workshop
Many scientific problem domains continue to be extremely data intensive. Traditional high performance computing (HPC) systems and the programming models for using them such as MPI were designed from a compute-centric perspective with an emphasis on achieving high floating point computation rates. But processing, memory, and storage technologies have not kept pace and there is a widening performance gap between computation and the data management infrastructure. Hence data management has become the performance bottleneck for a significant number of applications targeting HPC systems. Concurrently, there are increasing challenges in meeting the growing demand for analyzing experimental and observational data. In many cases, this is leading new communities to look towards HPC platforms. In addition, the broader computing space has seen a revolution in new tools and frameworks to support Big Data analysis and machine learning.
There is a growing need for convergence between these two worlds. Consequently, the U.S. Congressional Office of Management and Budget has informed the U.S. Department of Energy that new machines beyond the first exascale machines must address both the traditional simulation workloads as well as data intensive applications. This coming convergence prompted the integration of the PDSW and DISCS workshops into a single entity to address the common challenges.
The scope of PDSW is summarized as:
- Scalable storage architectures, archival storage, storage virtualization, emerging storage devices and techniques
- Performance benchmarking, resource management, and workload studies from production systems including both traditional HPC and data-intensive workloads.
- Programmability, APIs, and fault tolerance of storage systems
- Parallel file systems, metadata management, and complex data management, object and key-value storage, and other emerging data storage/retrieval techniques
- Programming models and frameworks for data intensive computing including extensions to traditional and nontraditional programming models, asynchronous multi-task programming models, or to data intensive programming models
- Techniques for data integrity, availability and reliability especially
- Productivity tools for data intensive computing, data mining and knowledge discovery
- Application or optimization of emerging “big data” frameworks towards scientific computing and analysis
- Techniques and architectures to enable cloud and container-based models for scientific computing and analysis
- Techniques for integrating compute into a complex memory and storage hierarchy facilitating in situ and in transit data processing
- Data filtering/compressing/reduction techniques that maintain sufficient scientific validity for large scale compute-intensive workloads
- Tools and techniques for managing data movement among compute and data intensive components both solely within the computational infrastructure as well as incorporating the memory/storage hierarchy
