Course Outline

Instructor:
Muthucumaru (``Mahes'') Maheswaran
McConnell Engineering Building, Room 213B
Phone: 398-1465
maheswar@cs.mcgill.ca

Office hours: Mon/Wed - 12:00noon - 2:00pm

Course Web site: http://www.cs.mcgill.ca/~maheswar/COURSES/ANC2004

Consultation: In person in the instructor's office during listed office hours or by appointment. Emails are welcome.

Course Description:

The immense popularity of the Internet has created a new class of distributed systems called network computing systems. Examples of systems that fall into this category include peer-to-peer systems, Grid computing systems, public computing utilities, and large-scale cluster computing systems. This course will examine the issues that are general to these systems and specific to each different system. By considering the issues in these perspectives the course will motivate the question "how should the Internet evolve to safely and efficiently deploy the next generation network computing systems?" What changes, if any, should be seen the protocols and end-node architectures.

The topics examined in this course can be grouped into three categories: fundamental concepts, enabling technologies, and system architectures and case studies. In fundamental concepts, following issues will be covered: characterizing network computing systems, network computing models, network computing design principles, scalability of network computing systems, and performance metrics and analysis of NC Systems. In enabling technologies, we cover fundamental technologies from distributed systems, resource and data dissemination and discovery, security and trust in network computing, and caching and replication. In system architectures and case studies, we cover intra-cluster resource management systems, Grid Computing Systems, peer-to-peer systems, and e-utility systems.

Pre-requisites: Courses in operating systems, computer networks, and very good Java/C programming skills.

Grading:

Class participation	15%
Paper reviews	20%
Mid term examination (open book exam)	15%
Final project (group of three or less)	50%

• Paper reviews: Before each class you are expected to read some paper/chapter of relevant material. This reading material will be assigned beforehand. The paper reviews (1 page) is due before the class. In the paper review you are expected to cover the following aspects with regard to the material covered in the paper: the major problem, the innovative method, significance in practical settings, possible improvements, possible ways of alternative performance evaluation. You will be graded based on the best 90% of the reviews (i.e., if we have 20 paper reviews, you need not submit 2 of them assuming the rest are perfect!).
• Class participation: For each class, there will be a facilitator who presents an overview of the material and poses relevant questions. The class is expected to engage in a discussion based on these questions. The facilitator will be graded on the quality of the overview presentation and the questions posed for the class. Each student is expected to be a facilitator for at least one class. You earn part of the grade under this category as the facilitator and part of the grade under this category by answering the questions (participating in the class). It is important you attend all the classes. If there is a need to be absent, please contact the Professor beforehand.
• Mid term exam: There will be one open-book exam. This will be held towards the end of the semester but before the finals period. The questions will be from the material discussed in the class. The specific scope of the exams will be defined well before the exam.
• Final project: You will be required to design and implement a substantial portion of a real network computing system. The objective is to get you intimately familiar with some of the emerging network computing technologies. The final project will be due in stages. There will be an initial project proposal with timelines, a mid-term progress report, and the final project (source code, documentation, report, and demonstration).

Course Topics

1. FUNDAMENTAL CONCEPTS
   • taxonomies for characterizing network computing systems
   • network computing design principles
   • performance metrics and analysis of NC Systems
   • scalability of network computing systems
2. ENABLING TECHNOLOGIES
   • naming
   • resource management algorithms
   • quality of service
   • performance prediction
   • security and trust
   • isolation and virtualization
   • discovery and dissemination
   • caching and replication
3. SYSTEM ARCHITECTURE AND CASE STUDIES
   • peer-to-peer systems
   • overlay networks
   • Grid computing systems
   • computing utilities

Book for part of the course:
Grid Resource Management (mirrored PDF)

References:
Grid Computing: Making the Global Infrastructure a Reality, Fran Berman, Geoffrey Fox, and Anthony Hey, John Wiley, 2004.

Academic Dishonesty: You are expected to abide by all the University regulations on academic dishonesty.
For the paper reviews: DO NOT just copy the abstract of the paper. It can be considered dishonesty! You are expected to read the papers carefully and write a review based on your own interpretation of the material presented in the paper.
For the final project: DO NOT share code with other groups. Receiving and providing code are equally WRONG! You are defeating the purpose of the learning experience that is supposed to be provided by this course.
When in doubt, consult the University guidelines or ask the Professor.