CS 386d Database Systems

This is a graduate-level introduction to the principles of database systems. We review and explain fundamental ideas and algorithms that are used in the construction of centralized DBMSs, distributed DBMSs, and database machines. Topics to be covered include:  query processing and optimization, database machines, object-oriented databases, concurrency control and recovery. Recent directions in database research are also surveyed.

Course Prerequisites

All students must have taken undergraduate database or its equivalent.

Lecture Notes, Texts, and Class Lectures

Lecture notes for the course will be made available at the Texas Union Copy Center. The first set of notes should now be available; additional sets of notes will be announced in class.  Class lectures are supplemented by papers (available below).  Students are responsible for reading these papers.  Remember: DO NOT PRINT THESE PAPERS ON CS PRINTERS!

• An excellent reference for concurrency control and recovery is by  Bernstein, Hadzilacos, & Goodman, Concurrency Control and Recovery.
• Additional postings on lecture notes, homework assignments, and software will be made available through this web page as they become available.

First two lectures on CC&R are here.

Presentations

The last class periods of the semester are devoted to student group presentations. These presentations will survey recent results presented in major database conferences. Group lectures will be critiqued prior to the actual presentation to enhance quality and content.  Topics for presentation can be chosen from any recent database conference, among them are:

• ACM SIGMOD
• Very Large Database Conference

Programming Projects and Homework

There will be a series of 3 programming projects during this course.  The first project will refresh your memory on SQL; the second will be either building an inverted file system using Berkeley DB (which is a file management system written in Java) OR a special project of your own. And the third project will be to develop a lock manager.  Details on these projects are forthcoming.

Project submission. To submit your project, execute on a CS Linux machine:

> turnin -submit <TBD> cs386d <file>

To check your submissions:

> turnin -list <TBD>  cs386d

Evaluation

Final grades will be determined approximately by the following scheme:

1. Your accumulative project grade will determine the maximum final grade for the course.  Ex: if you get a "B" average across all of your projects, your final grade will be no greater than a "B".
2. Final counts 40%, midterm counts 30% and presentation counts 30%.
   

Homework grades and class participation is used to decide borderline cases for final grades.

If you have difficulty meeting the requirements of this course, fail to hand in an assignment, or miss an exam because of extenuating circumstances, please advise the instructor in writing (not email) at the earliest possible date so that your situation can be discussed. If you encounter an unexpected medical or family emergency or a random act of Nature that causes you to miss the due date for homework or miss a quiz or exam, you must present suitable documentation in writing (not email) to the instructor before special consideration will be given. A file of all written correspondence will be kept by the instructor and decisions regarding them will be made at the end of the semester.

Numbers in [brackets] indicates the number of lecture on a topic. Copies of lecture notes are available at the Texas Union Copy Center, and the papers are available via the links below. Papers highlighted in yellow are subject for questions on exams. Homework assignments are indicated in orange.  And projects are in blue.  Please note that not all projects and their due dates have been posted on this page.

Query Processing [6]

• Basic Concepts in Centralized and Distributed Query Optimization [3]
  • Chaudhuri, "An Overview of Query Optimization in Relational Systems"
  • Selinger, et al., "Access Path Selection in a Relational Database Management System"
  • Mackert and Lohman, "R* Optimizer Validation and Performance Evaluation for Distributed Queries"
• Project #1: SQL Refresher
• Query Processing and Optimization in Database Machines[2]
  • Nyberg, Barclay, Cvetanovic, Gray, and Lomet, "AlphaSort: A Cache-Sensitive Parallel External Sort"
  • DeWitt, et. al., "The Gamma Database Machine Project"
• Object-Oriented Query Optimization
  • Hellerstein, "Predicate Migration: Optimizing Queries with Expensive Predicates"
• Problem Set #1
• Project #2: Berkeley DB DDL and Database Loading

Data Storage, Database Optimization, and XML [4]

• Spatial File Structures and Join Algorithms
  • Guttman, "R-Trees: A Dynamic Index Structure for Spatial Searching"
  • Hjaltason and Samet, "Incremental Distance Join Algorithms for Spatial Databases"
• Data Warehouses
  • Gray, et al., "Data Cube: A Relational Aggregation Operator Generalizing Group-by, Cross-Tab, and Sub-Totals"
  • Zhao, et al., "Simultaneous Optimization and Evaluation of Multiple Dimensional Queries"
• Gray Video, "The Revolution in Database Architecture"
  • Gray, "The Revolution in Database Architecture"
• Automatic Database Tuning
  • Zilio, et al. "DB2 Design Advisor: Integrated Automatic Physical Database Design"
• (Optional) XML databases and XQuery
  • Chamberlin, Robie, and Florescu, "Quilt: An XML Query Language for Heterogeneous Data Sources"
• Problem Set #2
• Project #3: Berkeley DB DDL and Database Loading
• Project #4: Berkeley DB DDL and Database Loading

Concurrency Control [7]

• Basic Concepts of Concurrency Control and Recovery
  • Bernstein text. "Serializability Theory"
• Problem Set #3
• Two-Phase Locking, Phantoms, and Multigranularity Locking [2]
  •  Mohan, "Interactions Between Query Optimization and Concurrency Control"
  • Bernstein text. "Two Phase Locking"
• Degrees of Consistency
  • Gray, et al. "Granularity of Locks and Degrees of Consistency in a Shared Data Base"
  • Batory, "Degrees of Consistency"
  • 
    
• Problem Set #4
• B+ Tree Locking Protocols
  • Lehman and Yao, "Efficient Locking for Concurrent Operations on B-Trees"
  • 
    
• Key Range Locking and Distributed Concurrency Control
  • Mohan, "ARIES/KVL: A Key-Value Locking Method for Concurrency Control of Multiaction Transactions Operating on B-Tree Indices"
• Non-Locking Protocols (Notes)
  • H.T. Kung and J.T. Robinson, "On Optimistic Methods for Concurrency Control"
• Problem Set #5

Recovery [3]

• Basic Concepts and Operation Logging [2]
  • Haerder and Reuter, "Principles of Transaction-Oriented Database Recovery"
  • Mohan, et al., "ARIES: A Transaction Recovery Method Supporting Fine-Granularity Locking and Partial Rollbacks Using Write-Ahead Logging"
• Recovery in Distributed Databases
  • Mohan, Lindsay, and Obermarck, "Transaction Management in the R* Distributed Database Management System"
• (Optional) Enterprise Java Beans

Current Trends in Database Research [4]

• Student Group Presentations [4]

Miscellaneous (Guest Lectures) [2]

• Conor Cunningham (Microsoft)

Revised: January 20, 2010.