CS 3420

Fault Tolerant Parallel and Distributed Systems

Fall 1997
Mondays and Wednesdays from 10:00AM to 11:20AM

113 MIB

Instructor

Rami Melhem (melhem@cs.pitt.edu)
219 Mineral Industries Building,
phone: 624-8426
Office Hours:
Mondays from 1:00 to 3:00
Thursdays from 1:00 to 3:00

Course Description

This seminar couse deals with the study of the principles of Fault Tolerant Computing. Topics covered include: Error detecting and correcting codes, hardware, software and time redundancy techniques, fault tolerant multiprocessors, system diagnosis and fault tolerance software, fault tolerance in distributed and real-time systems and performance and reliability evaluation techniques.

Reference text books (not required)

• Fault-Tolerant Computer System Design, by Dhiraj Pradhan - Prentice Hall.
• Fault Tolerance in Distributed Systems, by Pankaj Jalote - Prentice Hall.
  

Requirements and grading:

• One exam in October (15%).
• Paper presentations and class participation (25%).
• Class project and report (60%).

Course Outlines and Schedule:

• Presentations by the instructor
  Simple error detecting/correcting codes.
  Types of redundancy (hardware, software and time redundancies).
  Performance and reliability evaluation techniques.
  Fault tolerance in distributed systems.
  Fault tolerance in real-time systems.
  Fault tolerance in Multiprocessor systems.
• Presentations by students - possible topics:
  Using spare processor capacity for FT
  Reconfigureation/Embeddings in Multiprocessor systems
  FT communication and/or routing
  Roll-back and Forward recovery schemes
  Clock synchronization algorithms
  Group communication
  Common Object Request Broker Architecture (CORBA)
  Fault injection
  Formal verification methods and tools
  FT RT systems
• Presentations and Demonstrations of Class Projects
  to be determined

Course Index Department of Computer Science University of Pittsburgh