Welcome to the CSE232, Winter
99, home page
FINAL: Thursday 3:00pm, @APM4882, open books, notes,
computers, etc.
EXTRA OFFICE HOURS: Wednesday 10:30-11:00 and 12:10-3:00.
And then some more last minute office hours: Thursday
12:30-2:00.
CSE232 covers the implementation and operation of database
systems.
Lecture: TuTh 3:55-5:15, APM 5218
Instructor: Yannis Papakonstantinou, yannis@cs.ucsd.edu.
Guest Lectures by
Walt Burkhardt on RAID
Victor Vianu on semantic optimization
Class Web Page: Lecture notes, problems sets and
many announcements will be at the web site.A substantial part of the site
will be in postscript and pdf formats. Please configure
your browsers accordingly.
Prerequisites: CSE132A.
Texts: The "Database System Principles" by
Korth, Silberschatz and Sudarshan is recommended but not
required.
Exams: We'll have two midterms and one final. The
midterm dates will be announced soon.
Lecture Notes
- Jan 5: A quick overview of SQL and relational
algebra. Notes in html
and pdf.
- Jan 7: Course overview (html,
pdf)
- Jan 7: Physical aspects of disks and storage (html,
pdf)
- Jan 12: RAID Taxonomy: lecture by Prof.
Walt Burkhardt
- Jan 14: Indexing; elementary concepts and
algorithms (slides 1-12 from html,
pdf)
- Jan 19: Indexing; secondary indexes, B trees
(slides 13-30 from html,
pdf)
- Jan 21: B trees (con'td). Comparison with
conevntional indexes.
- Jan 26: Hashing indexes. Practice problems.
- Jan 28: Query Processing (html,
pdf)
- Feb 2: From SQL to Relational Algebra.
Algebraic Axioms. (html,
pdf)
- Feb 4: Algebraic Axioms (html,
pdf)
- Feb 9:
- Feb 11:
- Feb 16:
- Feb 18: Semantic Optimization I, by
Victor Vianu
- Feb 23: Midterm
- Feb 25: Semantic Optimization II, by
Victor Vianu
- Mar 3:
- Mar 5: Transaction Processing (html,
pdf)
Questions and Answers
- Readings on concurrency control and
recovery: The way the material is presented by
Korth & Silberschatz does not directly correspond to
my notes. Having said this, I recommend sections
13.1-13.6 and 14.1, 14.6 (not too much detail),
15.1-15.4.
- What is a blind write? How do we
determine that a write action is "blind"?
First, keep in mind that you cannot tell
whether a write is blind by looking only at the
transaction it appears in. You have to see the schedule
in which a write is used in order to call it blind.
Specifically, a write is blind if no transaction uses its
result and this result is also "erased" by a
subsequent write operation (i.e., this result does not
make it to the final state of the database)
Practice Problems