Community Affairs

Group Type: Research Group

Group Chair(s): Geoffrey Fox, christopher dabrowski

Group Email: gridrel-rg@ogf.org (Subscribe) (Archive by date) (Archive by thread)

This RG will assemble a community of standards developers and researchers to investigate requirements for reliability and robustness of future grid computing systems based on emerging Web Services and Grid standards. The primary purpose of the RG will be to develop an informational document that summarizes the state of current work on Grid system reliability and describes requirements for establishing and maintaining high levels of reliability in future large-scale Grids. An extended outline for this document is attached. This informational document will also provide preliminary requirements for methods and tools to measure Grid and WS system reliability. The informational document will be designed to impact Grid standard specifications currently under development. The RG will address questions on how these specifications might be changed to better enable large-scale grids to detect and overcome failures so that these systems can provide a level of robustness needed for industrial and scientific purposes.

Scope

The scope of this effort will center on improving understanding reliability and robustness [1] issues in Grid computing systems and on describing requirements for reliability of Grid systems developed on the basis of specifications of the GGF and related standards organizations. In investigating these issues and requirements, two considerations will merit special attention. First, the scale of grid computing systems is expected to grow dramatically as grid technology transitions to industrial use. Second, operational grid systems are likely to be subjected to volatile and uncertain conditions that potentially endanger or severely degrade their effectiveness in everyday use.
The effort will also include a survey of mechanisms employed by current Grid systems to respond to adverse circumstances to ensure system reliability and robustness. These mechanisms include, but are not limited to, Grid FTP, Grid monitoring services, Grid replication services, checkpointing and recovery services, autonomic computing services impacting grid system reliability, as well as mechanisms for maintaining consistent system and component states though time. The RG will consider the relationship of these mechanisms to, and interactions with, grid specifications being developed within GGF and other organizations. Current mechanisms, methods, and practices are expected to constitute a source of identified reliability and robustness issues and requirements.

Goals

1. The primary goal of the RG will be to develop an informational document that summarizes the state of current work on Grid and WS system reliability, identifies key issues, and describes requirements for establishing and maintaining high levels of reliability in future large-scale Grids. This informational document will also provide preliminary requirements for methods and tools to measure Grid and WS system reliability. The informational document will reflect the general experiences and findings of Grid system practitioners on grid systems reliability. An important goal of this document will be to serve as a guide on reliability issues for GGF working groups developing specifications, where appropriate.
2. To understand Grid reliability issues and requirements, the RG will organize forums and workshops for researchers, application developers, and others to present results of their work on reliability and robustness of grid systems and to exchange information. The RG may also solicit and obtain information on reliability requirements through web pages, mailing lists, white papers, best practices documents, and other publications that discuss:
� Standard specifications of the GGF and those of related organizations, including standards committees focusing on web-services used in grid systems.
� Grid services and software tools.
� Real-world Grid usage.
3. As a secondary goal, to promote and facilitate:
� Collaborations between researchers in grid systems reliability and robustness.
� Access to test beds and simulation models that investigate reliability issues
� Development of testing products, metrics, and evaluation activities.
4. Also as a secondary goal, the RG will also encourage and support research for developing test methods and metrics for evaluating grid systems reliability and robustness. This includes methods and metrics for evaluating the ability of grid systems to detect, and respond to, various kinds of failures, such as failures of individual components, links, as well as entire subnetworks. This also includes techniques (languages, terminology, tools) for risk assessment and evaluation of grid reliability relevant for industrial use. The RG would foster definition of benchmarks and minimum performance levels or thresholds for grid system reliability and robustness. Other issues of valid concern include, but are not limited to, evaluation of the stability of service interface versions (for grid and web service) that may differ across a network, whose interactions may result in unpredicted instability.

Deliverables and Milestones

1. A GGF informational document will be produced on reliability and robustness in Grid computing systems. This document is described above. An extended outline for this document is attached.
Timeframe: finalized in December 2007. An interim draft will be available for review by February 2007.
2. One or more workshop will be held, as needed, at future GGF meetings during which participants report results of research on reliability and robustness in grid systems. The workshops may also identify reliability and robustness issues within specifications being produced by GGF and related organizations and collect requirements and case studies.
Timeframe: Held on as needed basis.

View Group Charter

Group's Gridforge Project