Lecture Summary
ECE 428/CS 425/CSE 424 Distributed Systems, Fall 2005
Chapter/section numbers below refer to that from 4th edition of the textbook,
unless otherwise specified -- if you are using third edition of the textbook,
you can identify relevant chapters by comparing the tables of contents for
the two editions, which can be accessed from the textbook web links provided
on homepage for this course.
Readings listed below
in the entry for a lecture may actually correspond to multiple lectures
in the vicinity of that lecture. Not all topics listed for reading may be
covered in class. Some readings may be assigned ahead of the corresponding
lectures.
-
8/25/05 (lecture by Prof. Yih-Chun Hu): Course handout distributed.
Introduction to the course.
(slides)
Reading: Chapter 1
-
8/30/05: Programming assignment 0. Clock synchronization.
Synchronization in synchronous systems. Uncertainty (u) in message
delay. For n processes, optimum bound on clock skew is u(1 - 1/n).
Proof that, for 2 processes, the optimum bound on skew is u/2.
Cristian's algorithm.
(slides)
Reading: Sections 11.1 through
11.5
from chapter 11 (4th edition) on Time and Global States (or Sections 10.1 through 10.5 from 3rd edition)
-
9/1/05: Berkely algorithm. NTP. Logical clocks. Happened before relation.
Lamport clock. Vector clocks.
(slides)
Reading:
Sections 11.1 through 11.5
from chapter 11 (4th edition) on Time and Global States (or Sections 10.1 throug
h 10.5 from 3rd edition)
-
9/6/05: Global states. Consistent and inconcistent cuts. Chandy and Lamport's
"snapshot" algorithm (slides)
-
9/8/05: Consensus in distributed systems
Reading:
Section 12.5 of 4th edition (or Section 11.5 from 3rd edition).
Sections 1, 2 and 3 of the paper
Impossibility of Distributed Consensus with One Faulty Process,
Fischer, Lynch and Paterson, Journal of the ACM, April 1985.
-
9/13/05: Consensus in distributed systems. Algorithm OM(m) for
Byzantine agreement with m failures. Proof of FLP result for asynchronous
systems. (slides)
Reading: OM(m) algorithm from Section 3 of the paper
The Byzantine Generals Problem by Lamport, Shostak, Pease, in ACM
Transactions on Programming Languages and Systems, 1982.
-
9/15/05: Consensus in distributed systems.
Proof of FLP result for asynchronous systems.
Consensus in synchronous systems with up to f
crash-stop failures. Mutual exclusion.
(slide set 1 slide set 2)
Reading: Section 11.2 on Mutual Exclusion from 4th edition (or 10.2 of 3rd edition)
-
9/20/05:
Mutual exclusion.
(slides)
Reading: Section 12.2 on Mutual Exclusion from 4th edition (or 11.2 of 3rd edition)
-
9/22/05:
Multicast.
(slides)
Reading: Section 12.4 on Multicast from 4th edition (or 11.4 of 3rd edition)
-
9/27/05:
Elections.
(slides)
Readings: Section 12.3 on Elections from 4th edition (or 11.3 of 3rd edition)
-
9/29/05: Elections (Bully algorthm). Determining maximum recoverable state
with independent checkpointing and message logging.
Readings: Sections 1.1, 1.2, and 2.1 (you can omit
the proofs) of
Distributed System Fault Tolerance Using Message
Logging and Checkpointing,
David Johnson, Ph.D. thesis, Rice University, December 1989.
-
10/4/05:
State interval index, dependency vector, system state, optimistic and
pessimistic message logging, checkpointing, rollbak recovery,
consistent state,
determining maximum recoverable state.
Failure detector.
(slides)
Readings:
See readings in previous lecture (from Johnson's thesis)
Sections 2.3.2,
Section 11.1 (4th edition) or 10.1 (3rd edition).
-
10/6/05: Networking. (slide set 1,
slide set 2)
Readings: Sections 3.1 through 3.4 from Chapter 3 (Networking and Internetworking)
-
10/11/05: Peer-to-peer systems. (slide set 1,
slide set 2)
Readings:
Gnutella Protocol Specification,
SIGCOMM paper on Chord (Sections 1 through 4: You may omit the
proofs)
-
10/13/05: Peer-to-peer systems (Chord). Dynamic source routing
in ad hoc networks.
-
10/18/05 (Prof. Yih-Chun Hu): Security
Readings: Chapter 7 (Security). Lecture notes (PDF file)
br>
-
10/20/05: Mid-term test (in class)
-
10/25/05 (Prof. Yih-Chun Hu): Security
Readings: Chapter 7 (Security)
-
10/27/05: Dynamic source routing (DSR) in ad hoc networks.
Weak duplicate address detection (weak DAD) in ad hoc networks.
Transactions.
Slides for DSR (see relevant slides describing DSR protocol), slides for
weak DAD
slides for transactions
Readings:
-
relevant paper on DSR to be provided here
-
Weak
Duplicate Address Detection in Mobile Ad Hoc Networks,
N. H. Vaidya, ACM International Symposium on Mobile Ad Hoc Networking and
Computing (MobiHoc), June 2002.
-
Sections 13.1 through 13.3 of 4th edition - from chapter
on Transactions
(Sections 12.1 through 12.3 of 3rd edition)
-
11/1/05: Transactions. Concurrency control (slides
set 1,
set 2)
Readings: Sections 13.4-13.7 of 4th edition (or 12.4-12.7 of 3rd edition)
Chapter 14 of 4th edition (or chapter 13 of 3rd edition)
-
11/3/05: Concurrency control.
-
11/8/05 (Charles Yang): RPCs and Distributed Objects
(slides)
-
11/10/05: Concurrency control.
-
11/15/05: Distributed transactions
(slides).
Readings: Chapter 14 of 4th edition (or chapter 13 of 3rd edition)
-
11/17/05: Replication control
(slide set 1).
Readings: Sections 15.1-15.3 of 4th edition (or sections 14.1-14.3 of 3rd edition)
-
11/29/05 (Prof. Indranil Gupta): Replication control (slide set 2)
Readings: Chapter 15.4 of 4th edition (or chapter 14.4 of 3rd edition)
-
12/1/05 (Prof. Yih-Chun Hu):
Distributed file systems (slides)
Readings: Chapter 8 on Distributed File Systems
-
12/6/05:
Distributed shared memory (slides)
Readings: Chapter 18 on Distributed Shared Memory from 4th edition
(or chapter 16 from 3rd edition), and the slides above
-
12/8/05 (last lecture for the course):
Transactions and replication (slides)
Readings: Chapter 15.5 of 4th edition (or chapter 14.5 of 3rd edition) --
only Quorum Consensus Methods (section 15.5.5 from 4th edition,
and section 14.5.5 from 3rd edition) required for the final exam.
Return to the home
page for ECE 428