CS-15-829 D Advanced Topics in Software Systems: Data Stream Processing Fall 2003 Instructor: Christopher Olston MW 10:30-11:50, NSH 3002

Overview: In this course we will study the emerging research area of data stream processing, which is currently receiving a great deal of attention within the data management community. The focus is on online monitoring applications, in which continuous queries operate in near real-time over rapid and unbounded "streams" of data such as telephone call records, sensor readings, web usage logs, network packet traces, etc. We will cover a range of papers on this topic, from theoretical results to descriptions of prototype implementations. One of the objectives of the course will be to understand the fundamental differences between data stream processing and traditional data management.

Prerequesites: 15-415 (or equivalent) required; knowledge of relational database implementation techniques (i.e. 15-721 material) strongly recommended

Course meeting time/place: MW 10:30-11:50, NSH 3002

Office hours: MW 1-2, WeH 8206 (and by appointment)

Instructor email:

Schedule & Reading List

Grading

• Paper write-ups = 25%
• In-class presentations = 25%
• Project = 50%

Assignments

• Write-ups

  • Please compose a short write-up for each required paper from the reading list. (You are exempt from the write-up for papers you present in class.)
  • You may email your write-ups to the instructor with subject line prefix "[15-829D write-up]". Typed hardcopy is also accepted.
  • Write-ups are due by the beginning of class.

   

• Presentations

  • Students are required to present papers from the reading list in class.
  • Sign-ups will be arranged on the first day of class.
  • IMPORTANT: Please schedule a half-hour meeting with the instructor prior to the presentation to discuss what will be covered. Preferred meeting times: For Monday presentations, previous Thursday at 1:30. For Wednesday presentations, Tuesday (day before) at 2:00.
  • In some cases, if you ask nicely you can obtain slides from one of the student authors of the paper.

   

• Projects

  • Students are asked to complete an open-ended implementation project as part of this course.
  • Projects should be research-oriented. Simply implementing an algorithm from one of the papers is not sufficient. Click here for some hints, guidelines, and starting points.
  • Please consult instructor by October 15 for approval of your project idea.
  • Projects may be performed individually, or in groups of 2-3. (Project difficulty should be commensurate with size of group.)
  • Students will demonstrate their implementations in class at the end of the semester. If at all possible, demonstrations should run over real data. (Don't worry too much about the UI, provided the audience can understand what's going on.)
  • A paper describing the project, including an overview of the implementation and discussion of research outcomes is due at the end of the semester. Aim for a 15-20 page report.
  • NEW: demonstrations will occur on December 1 and December 3, in class. The schedule is as follows:
    • Monday December 1: Sandeep; Jimeng + Tim
    • Wednesday December 3: Deepay; Vlad + Amit
  • NEW: final project reports are due Friday, December 5th at 5pm eastern time.