CSE225 Grids and High Performance Distributed Computing

Spring 2004

WedFri5-620pm, HSS 2305B

First meeting:  Wednesday, March 31, 2004

 

Professor:   Andrew A. Chien achien@ucsd.edu APM 4808

 

CSE225 Materials

 

Course Overview and Outline

 

Computational and Data Grids are open, shared resource infrastructures which support high-capability dynamic distributed applications.  These infrastructures are the focus of national research projects in many countries (USA, UK, Japan, Korea, India, etc.), and the commercial strategies of major computing companies (IBM, Sun, Hewlett-Packard, etc.). 

 

For Grids to realize their potential, robust solutions to a number of difficult problems must be developed.   In CSE225, we will begin with an overview of the unique attributes of grids, then study model problems for these challenges.  These model problems include:

 

1.      Dynamic Applications are Resource Aware: “Beauty is in the eye of the beholder.”

a.      Resource Description and Selection (single node, aggregates, pipelines)

b.      Network and Application Uncertainty

c.      Dynamic monitoring and Adaptation/Reconfiguration

 

2.      Open Resource Sharing: “You can’t always get what you want.”

a.      Asymmetric Resource Sharing: Desktop Grids

b.      Resource Managers and Co-allocation/Reservation

c.      Best-effort Slice and Virtualization Systems

d.      Economic techniques and Stability

 

3.      Federated Security: “If you can’t trust the authorities, who can you trust.”

a.      Security Infrastructures and Grids

b.      Virtual Organizations, Community Approaches

c.      Web of Trust, Reputation Systems

d.      Unsecured Infrastructures: Application-level techniques

 

4.      Dynamic Distribution of Data-Intensive Applications: “Let the data decide.”

a.      Fixed:  Data: Distributed Databases, Web Services Model

b.      Mixed: EU DataGrid and OGSA-DAI

c.      Mobile Data: OptIPuter Storage Model

 

Students should complete the course well prepared to undertake research in Grids and High Performance Distributed Computing

 

Coursework will includes technical paper readings, written homeworks, lab exercises, and a course project in one of these areas.  Students will also gain hands-on experience with some infrastructure for Grid computing (possibilities include the Globus Toolkit, Jini, Condor, XtremeWeb, Boinc, etc.).

 

 

 

 

 CSAG Home Page

 

 UCSD CSE Home

 

  UCSD Home