MTAGS17: 10th Workshop on Many-Task Computing on Clouds, Grids, and Supercomputers
Denver, Colorado -- November 17th, 2017
The 10th workshop on Many-Task Computing on Clouds, Grids, and Supercomputers (MTAGS17) will provide both the scientific and industrial communities a dedicated forum for presenting new research, development, and deployment efforts of algorithms, frameworks, and systems for many-task computing (MTC), machine learning and big data applications on large scale clusters, clouds, grids, and supercomputers. The theme of the workshop encompasses loosely-coupled applications driven by big data. The applications are generally composed of many tasks (e.g., millions to billions) to achieve some larger application goal. This workshop will cover challenges that can hamper efficiency and utilization in running applications on extreme-scale systems, such as local resource manager’s scalability and granularity, data-aware scheduling, efficient utilization of intra-node parallelism, parallel file-system contention and scalability, data locality, I/O management, reliability at scale, and application scalability. We welcome paper submissions in theoretical, simulations, and real systems topics with special consideration to papers addressing the intersection of petascale/exascale challenges with large-scale cloud computing and machine learning. Papers will be peer-reviewed, and accepted papers will be published in the workshop proceedings as part of the ACM SIGHPC. The workshop will be held in conjunction with SC17---The International Conference on High Performance Computing, Networking, Storage and Analysis---in Denver, Colorado, USA.
The advent of computation can be compared, in terms of the breadth and depth of its impact on research and scholarship, to the invention of writing and the development of modern mathematics. Scientific Computing has already begun to change how science is done, enabling scientific breakthroughs through new kinds of experiments that would have been impossible only a decade ago. As computing becomes a pervasive part of the scientific process, there is a great opportunity to make powerful computing techniques, previously reserved for projects with only the largest investments, available to a broad scientific community.
The massive increase in intra-node
parallelism provided by modern hardware presents a challenge to both traditional
parallel scientific applications and loosely-coupled data driven applications
with large existing investments in previously developed software that are extremely
difficult to be redesigned from scratch using the latest programming models
(e.g., MapReduce, Spark). Many-task computing (MTC) studies technologies,
simple and advanced, to rapidly compose highly scalable applications from
existing sequential codes. MTC encompasses loosely-coupled applications, which
are generally composed of many tasks (both independent and dependent tasks) to
achieve some larger application goal. Growing from the successes of Globus,
Condor, and national-scale grid computing infrastructures, MTC techniques have
been deployed on many systems from single many-core systems (leveraging GPGPUs
and Intel’s Xeon-Phi chips), to the largest multi-petascale high-performance
computing (HPC) systems. The development and deployment of these MTC systems
have expanded the utility of the underlying technologies and fed back to improve
the performance and usability of the technologies themselves. Similarly,
technologies developed for cloud computing (including MapReduce-based and Spark
models) can provide additional connections and innovations in computing
techniques. MTAGS is a unique venue to
promote HPC-related concepts to the broader scientific and cloud computing
communities.
We are entering a “big data” era, as advances in networking, instrumentation, simulation technologies, Internet computing, and social networks are producing data at an unprecedented rate. The collection, storage, analysis, and sharing of this data are thus one of the greatest challenges in the 21st century. Advanced machine learning algorithms and technologies are demanded to analyze the data and extract meaningful information from the structured, unstructured, and semi-structured data. Many of these machine learning and big data applications inherit the loosely-coupled data driven nature of MTC paradigm. Support for machine learning and big data applications is critical to advancing both scientific and industrial domains as storage systems have experienced an increasing gap between its capacity and its bandwidth by more than 10-fold over the last decade. There is an emerging need for advanced techniques to manipulate, visualize, and interpret large datasets. While these problems are deemed as some of the most critical technical barriers to overcome in communities of cloud computing, machine learning, and big data, many of them have been or are being actively studied as well in HPC and MTC. Therefore, this workshop also provides a channel to exchange large-scale data management technologies between scientific applications and industrial techniques, which is another mission of MTAGS17.
The 10th workshop on Many-Task
Computing on Clouds, Grids, and Supercomputers (MTAGS17) will provide both the
scientific and industrial communities a dedicated forum for presenting new
research, development, and deployment efforts of algorithms, frameworks, and
systems for many-task computing (MTC), machine learning and big data
applications on large scale clusters, clouds, grids, and supercomputers. This workshop
encourages interaction and cross-pollination between those developing
applications, algorithms, software, hardware, and networking, emphasizing many-task
computing for large-scale distributed systems. We believe the workshop will render
an open platform to (re)define the current state-of-the-art architectures and
services for future high-end computing infrastructure.
For more information on past workshops, please see MTAGS16, MTAGS15, MTAGS14, MTAGS13, MTAGS12, MTAGS11, MTAGS10, MTAGS09, and MTAGS08. We also ran a special issue on Many-Task Computing in the IEEE Transactions on Parallel and Distributed Systems (TPDS) which appeared in June 2011, and it can be found at http://datasys.cs.iit.edu/events/TPDS_MTC; the proceedings can be found online at http://www.computer.org/portal/web/csdl/abs/trans/td/2011/06/ttd201106toc.htm. In addition, we have run a special issue on Many-Task Computing in the Cloud in the IEEE Transaction on Cloud Computing: http://datasys.cs.iit.edu/events/TCC-MTC15/. We, the workshop organizers, also published two papers that are highly relevant to this workshop. One paper is titled "Toward Loosely Coupled Programming on Petascale Systems", and was published in SC08; the second paper is titled “Many-Task Computing for Grids and Supercomputers”, which was published in MTAGS08, both of which have been highly cited, with 136 and 237 citations respectively.
We invite the submission of original work that is related to the topics below. The papers should be 8 pages, including all figures and references. We aim to cover topics related to Many-Task Computing on each of the three major distributed systems paradigms, Cloud Computing, Grid Computing and Supercomputing. Topics of interest include:
Compute resource management
o Distributed scheduling algorithm
o Runtime environment
o Intra-node resource management
o Performance evaluation of resource managers in use on large scale systems
o Dynamic resource allocation
o Power-aware scheduling
o Scheduling applications' executions with little performance variation between different runs
o Techniques to manage and schedule generic resources including MIC and/or GPUs
o Challenges and opportunities in scheduling and running loosely-coupled data-intensive applications on HPC systems and Cloud Computing infrastructures
Data storage architectures and implementations
o Parallel and distributed file systems
o Multi-tier data stores
o NoSQL data storage system
o Distributed meta-data management
o Data caching frameworks and techniques
o Data provenance
o Data management within and across data
centers
o Data-aware and locality-aware
scheduling
o Data-intensive computing applications
o Eventual-consistency storage usage and
management
Programming models and tools
o MapReduce/Hadoop/Spark and their generalizations and implementations
o Many-task computing and cloud computing middleware and applications
o Parallel inter-node programming frameworks
o Distributed intra-node programming models
o MPI fault tolerant techniques
o Scalable and reliable overlay network
o Ensemble MPI techniques and frameworks
o Service-oriented science applications
Large-scale workflow systems
o Workflow system performance and scalability analysis
o Scalability of workflow systems
o Parallel programming language
o Workflow infrastructure and e-Science middleware
o Application data dependency graph generation tool
Large-scale loosely coupled applications
o Data-intensive MTC applications
o High-throughput computing (HTC) applications
o Machine learning and big data applications
o Quasi-supercomputing applications, deployments, and experiences
Performance evaluation
o Real systems
o Discrete event simulations (DES) and Parallel discrete event simulations (PDES)
o Docker container techniques
o Reliability of large systems
o Application performance tuning
- Full paper due: August 27th, 2017
- Acceptance notification: September 29th, 2017
- Camera Ready Due: October 13th, 2017
- Workshop date: November 17th, 2017
Authors are invited to submit papers with unpublished, original work of not more than 8 pages of double column text using single spaced 10 point size on 8.5 x 11 inch pages, as per IEEE 8.5 x 11 manuscript guidelines; document templates can be found at http://www.ieee.org/conferences_events/conferences/publishing/templates.html. The final 8-page papers in PDF format must be submitted online at easychair before deadline: https://easychair.org/conferences/?conf=mtags17. Papers will be peer-reviewed for novelty, scientific merit, and scope for the workshop. Submission implies the willingness of at least one of the authors to register and present the paper.
General Chairs
- Ke Wang, Microsoft Corportation, USA
- Dongfang Zhao, University of Nevada, USA
Steering Committee
- Ioan Raicu, Illinois Institute of Technology, USA
- Justin Wozniak, university of Chicago & Argonne National Laboratory, USA
- Ian Foster, University of Chicago & Argonne National Laboratory
- Yong Zhao, University of Electronic Science and Technology of China
Program Committee
- Kyle Chard, University of Chicago, USA
- Evangelinos Constantinos, Massachusetts Institute of Technology, USA
- Bo Feng, Capital One, USA
- Florin Isaila, Universidad Carlos III de Madrid, Spain
- Jik-Soo Kim, KISTI, Korea
- Anthony Kougkas, Illinois Institute of Technology, USA
- Michael Lang, Los Alamos National Laboratory, USA
- Christopher Moretti, Princeton University, USA
- Bogdan Nicolae, IBM Research, Ireland
- David O'Hallaron, Carnegie Mellon University, USA
- Ana-Maria Oprescu, University of Amsterdam, Netherlands
- Judy Qiu, Indiana University, USA
- Matei Ripeanu, University of British Columbia, Canada
- Iman Sadooghi, Bank of America, USA
- Mike Wilde, University of Chicago & Argonne National Laboratory, USA
- Xu Yang, Amazon, USA
- Zhao Zhang, Texas Advanced Computing Center, USA
- Zhou Zhou, Salesforce, USA