Workshop on Service-Oriented Computing Performance: Aspects, Issues, and Approaches

http://www.cs.binghamton.edu/~kchiu/socp07/

Monterey Bay, California

June 26, 2007

In conjunction with HPDC 2007

Service-oriented computing (SOC) is an emerging paradigm that is changing the way systems are designed, architected, deployed, and used. SOC decomposes computation into a set of loosely-coupled, abstract services, and emphasizes document-centric interactions through the exchange of messages. Services can be composed, nested, and orchestrated into a variety of control patterns and workflows. SOC has seen adoption in areas such scientific computing, Grid computing, and business computing, and can facilitate wide-scale application integration within and across organizational boundaries.

SOC’s loosely-coupled, document-centricity, and high degrees of encapsulation and self-description challenge performance in a number of aspects. New techniques of performance analysis, modeling, and prediction can address some of these challenges, but further research is still needed. Different programming paradigms, design methodologies, or programming language principles also may reduce or eliminate some of the abstraction, encapsulation, and composition costs of SOC. Multicore chips and cluster-wide parallelism also offer interesting avenues for improving and investigating SOC performance. Advanced processing techniques or encodings for languages such as XML also may play a role.

We invite innovative papers on any aspect of performance and SOC from all communities, such as the WWW community, the programming languages community, and the Grid community. We welcome different types of papers, including experimental, works-in-progress, and position papers. By bringing together different communities, perspectives, and approaches, this workshop will seek to focus and clarify the state-of-the-art, leading to cross-fertilization. Topics include, but are not limited to:

• Distributed, multicore, and parallel processing and protocols (execution models, architectures, properties, performance evaluation)
• Alternative XML representations and encodings, such as “binary” XML
• Code generation, incremental, lazy, and streaming techniques for improving the performance of SOC at all levels of the services stack
• Programming language and compiler techniques such type-based optimizations, static analysis, and transformations
• Hardware acceleration techniques for SOC such as reconfigurable computing and hybrid computing
• Reducing encapsulation, abstraction, orchestration, and composition costs of SOC
• Performance analysis, modeling, and prediction as it relates to SOC
• Security and performance

Program

Opening Remarks

8:50-9:00

Session 1

9:00-10:15

ISC: Providing Efficient XML-Based Service-Orientation for Core OS Functionality

Hermann Schloss, University of Trier; Ingo Scholtes, University of Trier; Peter Sturm, University of Trier

9:00-9:25

McGrid: Framework for Optimizing Grid Middleware on Multi-Core Processors

Rajdeep Bhowmik, SUNY Binghamton; Chaitali Gupta, SUNY Binghamton; Madhusudhan Govindaraju, SUNY Binghamton; Aneesh Aggarwal, SUNY Binghamton

9:25-9:50

Comparing the Use of Bayesian Networks and Neural Networks in Response Time Modeling for Service-Oriented Systems

Rui Zhang, University of Oxford; Alan Bivens, IBM Research

9:50-10:15

Coffee Break

10:15-10:45

Session 2

10:45-12:00

Parallel XML Processing by Work Stealing

Wei Lu, Indiana University; Dennis Gannon, Indiana University

10:45-11:10

Benefits of Alternate XML Serialization Formats in Scientific Computing

Jaakko Kangasharju, Helsinki Institute for Information Technology; Sasu Tarkoma, Helsinki Institute for Information Technology

11:10-11:35

Approaching a Parallelized XML Parser Optimized for Multi-Core Processors

Michael Head, SUNY Binghamton; Madhusudhan Govindaraju, SUNY Binghamton

11:35-12:00

Lunch

12:00-1:30

Session 3

1:30-2:45

A QoS-Based Selection Approach of Autonomic Grid Services

Jonatha Anselmi, Politecnico di Milano; Danilo Ardagna, Politecnico di Milano; Paolo Cremonesi, Politecnico di Milano

1:30-1:55

Empowering Distributed Workflow with the Data Capacitor: Maximizing Lustre Performance across the Wide Area Network

Stephen C. Simms, Indiana University; Gregory G. Pike, Oak Ridge National Laboratory; S. Teige, Indiana University; Yu Ma, Indiana University; Bret Hammond, Indiana University; Larry L. Simms, Indiana University; Doug Balog, Pittsburgh Supercomputing Center

1:55-2:20

Data Transfer Performance Issues for a Web Services Interface to Synchrotron Experiments

Paul D. Coddington, University of Adelaide; Andrew L. Wendelborn, Univserity of Adelaide; Donglai Zhang, University of Adelaide

2:20-2:45

Not Presented

Comparing Semantic Registries - OWLJessKB and InstanceStore

Simone A. Ludwig, University of Saskatchewan; Omer F. Rana, Cardiff University

Organization

Organizers

Program Committee

Contact