Petrenko A.I., Svistunov S.Ya., Kiselev G.D. Grid Technologies Practical Course

Petrenko. Grid Petrenko A.I., Svistunov S.Ya., Kiselev G.D.
Grid Technologies Practical Course. - Kyiv. NTUU "KPI", 2011. - 448 p.

The goal of this practical course on grid technologies is to introduce the concept and the current state of the art in grid computing to students, and to gain the skills of application of grid technologies for engineering analysis, optimization and design by students. It is designed for master's training at the new "System Design" specialty, but can also be useful for other specialties of "Computer Science" and "Computer Engineering" branches. The course can be recommended for students and graduate students of all IT-disciplines, as well as for their teachers.

This course contains necessary theoretical background and demonstration examples; it plays the role of educational and methodical support of the discipline "Grid Technologies for distributed computing and data processing". It is designated for acceleration of grid specialists' training, teaching them to work with novel tools for numerical calculations demanding high performance computing resources.

It consists of the cycle of experimental researches, aimed at the achievement of practical skills of using the grid technologies to solve scientific and practical problems:
- working with PBS-controlled computing cluster to learn the remote access to multiprocessor computing system and to gain practical skills of compiling and running the simple programs using the cluster management system;
- obtaining the user certificate: user learns to generate key pair (public key as the part of the certificate and private key for certificate signing), to obtain the certificate, to generate and use the proxy-certificate;
- learning the remote access to grid resources (operating under ARC middleware ): user uses UNIX-compatible computing cluster to design grid job xRSL description, runs job from the access server via ARC middleware with his proxy-certificate, and checks the running job status;
- learning the remote access to grid resources (operating under gLite middleware): user composes grid job JDL description, runs job from access server via gLite middleware with his proxy-certificate, checks the running job status, retrieves the results and analyzes them;
- learning the architecture of information and monitoring systems of production grids running under ARC and gLite middleware that collect and manage grid state data retrieved from numerous distributed sources-providers. These subsystems are responsible for permanent control of grid operation and for the instant reaction to the emerging errors;
- learning the providing grid access through the web service.

Thereby this practical course makes possible to obtain grid certificate, to prepare gird jobs for execution, to choose the target computing element, to control the execution process, to view the job execution results and to save these results on user's machine. It is expected that students will be able to gain practical experience of using the computing powers of NTUU "KPI", ITP and other connected supercomputers, as well as informational resources of the International Data Center "Geoinformatics and Sustainable Development".
Appendix: basic Linux OS commands (with Midnight Commander and vi/vim reference), ARC user interface commands and options, xRSL (Extended Resource Specification Language) attributes, gLite 3.1 command line interface reference, general structure of the JDL(Job Description Language).

Ukrainian (UA)Russian (CIS)English (United Kingdom)