Date |
HTML file |
Powerpoint file |
Topic |
Aug. 24 |
Introduction to the course, answers to common questions |
||
Aug. 29 |
Issues of Internet reliability. Basics of how IP and UDP work. Why the Web isn't reliable. End-to-End argument. |
||
Aug. 31 |
Everything anyone ever wanted to know about remote procedure call, and local optimizations for it. U-Net architecture. |
||
Sept. 5 |
The TCP sliding window protocol and optimizations. Why The Internet is really the bottom half of TCP |
||
Sept. 7 |
IP multicast, how multicast groups are formed, reliability and "good citizen" issues |
||
Sept. 12 |
Client-server computing, stateful and stateless architectures, MOMS, issues of server scalability |
||
Sept. 14 |
Transactions: how they work on non-distributed systems. 2PL. Extending to distributed systems (failures, 2PC) |
||
Sept. 19 |
Transactions: new ideas for transactional languages and transactions on objects. Nested transactions. Availability. |
||
Sept. 21 |
Replication and multicast problem. 2PC, 3PC and issues of failure detection |
||
Sept. 26 |
Agreement on membership as a substitute for reliable failure detection |
||
Sept. 28 |
Reliable, ordered multicast and virtual synchrony model |
||
Oct. 3 |
Using virtual synchrony: replicated data, state transfer, single and multigroup perspectives. Lightweight group motivation and concept |
||
Oct. 5 |
Relating models of consistency to cost and availability. Comparing transactional replication with virtual synchrony. |
||
Oct. 10 |
|
|
Fall break: October 7 - October 11 |
Oct. 12 |
Guest lecture: Robbert van Renesse. System architecture considerations. Building systems with layers to enhance flexibility. Optimizations to keep performance high and costs low. Dynamically switching stacks. |
||
Oct. 17 |
How and why systems fail. Can a replicated service be more reliable than the underlying servers? |
||
Oct. 19 |
Everything you always wanted to know about CORBA, COM and OO technologies |
||
Oct. 24 |
Internet QoS issues: what does quality of service mean? IPv6 features. |
||
Oct. 26 |
Farnam Jahanian (University of Michigan, Ann Arbor): From his lecture on scalability issues in the Internet, measuring Internet behavior |
||
Oct. 31 |
Internet QoS issues: RSVP, Diffsrv and limitations |
||
Nov. 2 |
Internet QoS as a reliability issue for Internet multicast: SRM, RMTP-II. Scaling issues in each. PGM, LGM, IBM Gryphon routers. Cisco routers. |
||
Nov. 7 |
Bimodal multicast (pbcast) |
||
Nov. 9 |
Next generation internet: Will it be "unsafe at any speed"? |
||
Nov. 14 |
Real-time systems: basic ideas. Clock synchronization. CASD multicast |
||
Nov. 16 |
Real-time systems: Friedman's experiments on cluster scalability |
||
Nov. 21 |
Lecture_25 |
Guest lecture: Werner Vogels. Galaxy project. Large-scale clusters |
|
Nov. 23 |
|
|
No class: Thanksgiving Recess Nov. 22-Nov. 27 |
Nov. 28 |
Scalability: how can we figure out if a technology will be adequately scalable? |
||
Nov. 30 |
Lecture_27 |
Future of distributed computing |