Grid Computing, NGSSC, 2p
Erik Elmroth, Olle Mulmo, and Leif Nixon
August 19, 2004
This course is designed to give a broad overview of the concept of
grid computing and state-of-the-art grid software, understanding and
practical experience of how computational grids can be used to further
research in science and technology, and to review some current and
future directions of grid computing research and development.
The prerequisities are extensive programming experience (e.g., in
C/C++ and Java) and completion of the following courses (or similar
knowledge and experience): Object-oriented scientific programming
NGSSC, 3p, Introduction to High Performance Computing NGSSC, 4p,
Parallel algorithms with applications to scientific computing, 3p,
Scientific visualization NGSSC, 3p.
Topics covered in the course include the following:
- Introduction to grid computing ( 10% of lectures):
- Overview: Fundamental concepts, grid architecture models, grid
middleware functionalities, etc.
- International overview: Application projects, software development
projects, political and administrative aspects on grids, emerging
grid standards (OGSA/OGSI), etc.
- Foundations in algorithm and software development for grids.
- Grid computing middleware and tools ( 40% of lectures):
- Globus: Overview and some specific components, including resource
specification language, security, scheduling, file transfer,
information services, etc.
- Grid security: Grid security demands and solutions for, e.g.,
authentication, authority, assurance, accounting, trust, group
communication, for large-scale, dynamic, multi-organization
- Resource management and scheduling: Characterization of resource
management problems based on job requirements, characteristics,
and availability of resources. Algorithms, tools and sample
resource management systems.
- Grid portals: Convenient access to grid environments.
Functionality and underlying infrastructure for sample general and
application specific portals.
- Data management: Key issues for data management in grids,
including file transfer, data replication, data caching issues,
- Sample grid middleware packages, software tools, and problem
solving environments for grids.
- Applications, algorithms, and sample projects ( 40% of lectures):
- Swegrid and NorduGrid ARC: Introduction to the Swegrid resources
and the NorduGrid software package. Introduction to hands-on
- Swegrid applications: Overview of planned and ongoing projects, in
e.g., high-energy physics, biomedical sciences, earth sciences,
space and astro sciences etc. Key issues for their successful use
of Swegrid and the grid's implications on future research in these
- Algorithm development for grids, and e.g., the GrADS project.
Focus on the special requirements that grid environments put on
algorithm development, e.g., heterogeneity in computers and
networks, fault tolerance, possibilities to access special
instruments and equipment, etc.
- Environments for visualization, interactivity and collaboration on
the grid: Overview and examples from computational steering, data
mining, virtual reality.
- Sample grid research projects ( 10% of lectures):
- One or two invited presentations on current grid research topics.
- Computer projects (preliminary):
- Hands-on grid introduction. Sample small ready-to-run exercises.
- Basic resource brokering.
- Hands-on introduction to OGSA/WSRF programming.
The course literature will be based on a collection of articles from
journals, conference proceedings and books, and material available via
WWW. Lecture notes including exercises, projects etc will be produced
for the course and made available via WWW.
A number of carefully documented computer assignments.
The course 2005 is planned to be given at NSC, Linköping University,
January 10-18, including scheduled and unscheduled work during the
weekend, January 15-16. As a guideline, the 80 hours of the two credit
points course will roughly be spent on 20 hours with lectures, 45
hours scheduled computer assignments (including introductions and
summaries) 15 hours homework (literature studies, unscheduled work on
The following years the course is planned to alternate between HPC2N
(Umeå University), NSC (Linköping University), and PDC (KTH).