Course Description

A computational grid is a hardware and software infrastructure for wide-area distributed computing. In a grid, computational, storage, and network resources are heterogeneous and reside in different security domains. This makes the computational grid environment much more challenging than traditional distributed computing environments, which typically consist of homogeneous machines under the control of a distributed operating system.

In this class, we discuss the basic services that must be provided by a computational grid infrastructure to enable high-performance applications to be distributed over the wide area. We survey recent technical papers on grids and focus particularly on middleware from the Globus project, the most successful grid infrastructure.

Course Topics:

1. Security for Wide Area Environments
-Authentication
-Authorization and access control
-Encryption

2. Resource Management
-Remote execution
-Scheduling
-Resource reservation and quality of service

3. Information Management
-Registering and locating grid resources
-Performance measurement: choosing the "best" resources

4. Data Management
-Managing petabytes of data
-Efficient, parallel transfers over wide area networks
-Caching and replication

5. Advanced Topics
-Including other grid projects, such as Condor, Legion, and the Storage Resource Broker

Required Prerequisites: Courses in operating systems and networks (preferably at the graduate level)