1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad...

13
1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of California, Irvine

Transcript of 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad...

Page 1: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

1

ICS 223: Transaction Processing and Distributed Data ManagementWinter 2011

Professor Sharad Mehrotra

Information and Computer Science

University of California, Irvine

Page 2: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

Course Outline• Introduction (1 class)• Theoretical Underpinnings (2 classes )

– transactions, schedules, conflict, view, and final-state serializability correctness beyond serializability

• Concurrency Control Mechanisms (2 classes)– -2PL, TO, validation protocols, SGT, Performance modeling– Deadlock detection, avoidance, lock conversion, update locking, etc.

• Exploiting structure & semantics to improve concurrency (2 classes)– path protocol, tree protocols, crabbing– operator semantics beyond R/W model -- escrowing– multiversioning

• Concurrency control in dynamic setting (2 classes)– phantoms, predicate locking, granule concept, MGL, key range locking,

dynamic KRL

• Mid Term Exam (1 class)

ICS214B Notes 01 2

Page 3: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

Course Outline• Recovery basics (3 classes)

– Shadowing, logging, multilevel recovery, ARIES, Checkpointing

• Distributed commit protocols (2 classes)– 2-phase, 3-phase, presumed nothing, presumed abort, presumed

commit– Various incarnations – distributed, centralized, hierarchical.

• Distributed Database Issues (2 classes)– Replicaiton, - laxy, eager, primary copy versus update anywhere, – Disaster recovery systems - 1-safe, 2-safe, optimized 1-safe

• Advanced transaction models (1 class)– Savepoints, nested transactions, sagas,

• Transactions and clouds (1 class) • Other topics (1 class)

– Transactions in NoSQL system, PNUTS, main-memory databases

ICS214B Notes 01 3

Page 4: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

Course Requirements

• Assignments -- 10%• Mid Term Exam -- 25%

– date TBA

• Final Exam -- 35%– check catalog for date

• Project --- 30%

Page 5: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

Assignment Policies• OK to do in groups (no more than 3) • Assignments must be submitted before class

on the assignment due date:– 1 copy per group

• Late submission:– No grace period after due date except under exceptional

circumstances– job interviews, out of town trip etc do not qualify as exceptional

circumstances!

Page 6: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

Projects – groups of size 3• Project Goals:

– serve as a vehicle learn beyond what is covered in class.

• Two types of projects:– Survey Oriented Projects

In depth exploration of a more current topic (e.g., transactions on the cloud, weak consistency models, transactions on flash storage, etc.)

– Research Oriented Projects in-depth exploration of a novel research idea

• Project Milestones and Deliverables:– Project Proposal (due in 3rd Week) – 1 page– Project Report (due at end of course) – 7 pages (12 pages for

survey projects) – Project Demonstration (last class) – 10 minute presentation– Project Presentation (to be scheduled) – 30 minute presentation

• List of project ideas will be posted on the web site before the end of the week.

Page 7: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

Key to successful project– START EARLY– CHOOSE GROUP MEMBERS WISELY (complement each

others skills)– START EARLY!!!

Page 8: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

Prerequisites

• Undergraduate course in database management (CS 184 or equivalent)– database design, relational model, SQL, object-oriented data

model, database design– You may brush up by reading up on Chapters 1-9 of the text

book

• Some operating systems concepts – virtual memory, paging, concurrent programming, semaphores,

critical sections, monitors, file and buffer management.

• Basic Computer Science Concepts:– depth first search, directed undirected graphs, big O notation,

computational complexity,

Page 9: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

ICS214B Notes 01 9

Course General Info

• URL: https://sites.google.com/site/cs223a/

• Lecture times: Mon-Wed 2 – 3.20pm

• Instructor: Sharad Mehrotra, [email protected]

• Grader: Kerim Y. Oktay

[email protected]

Page 10: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

ICS214B Notes 01 10

Text BooksNo required textbooks. The following are recommended:• Database System Implementation, by Hector Garcia-

Molina, Jeffrey Ullman, and Jennifer Widom, Prentice Hall, 1999. Or Database Systems: The Complete Book, by Hector Garcia-Molina, Jeffrey D. Ullman, Jennifer D. Widom, Prentice Hall, 2001.

• Database Systems concepts, Silberschatz, Korth, Sudarshan

• Some of the lecture material derived from– Transaction Processing: Concepts and Techniques Jim Gray and

Andreas Reuter Morgan Kaufman Publishers 1993

• Research papers may be made available whenever possible to supplement the textbook.

Page 11: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

ICS214B Notes 01 11

Grading• To pass the course (B or above), you MUST score:

– 75% or above in the assignments

– 60% or above in the mid term and finals.

– 75% or above in the project least grade to pass the course = 66%

• To get a guaranteed A– Meet the passing criteria

– Get over 85% overall

Page 12: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

ICS214B Notes 01 12

Simplified DBMS Architecture

Database andIndices

TransactionManager

Buffer manager

File system

Metadataand data

dictionary

compilers

evaluatoroptimizer Query processor

Storage manager

Application Queries Schema changes

Page 13: 1 ICS 223: Transaction Processing and Distributed Data Management Winter 2011 Professor Sharad Mehrotra Information and Computer Science University of.

ICS214B Notes 01 13