Distributed Systems
Corso di Laurea Specialistica in Ingegneria Informatica
AA 2007/2008
ORDINAMENTO 2000 (5 Credits)
Prof. Roberto Baldoni
Exam 19/3/2008 - Marks
Thursday 4.30pm - 6pm. Via Ariosto 25 - room B212
| Wednesday | Room A3 (Via Ariosto) |
3.45pm-5.15pm, 5.30pm-7.00pm |
| Thursday | Room A3 (Via Ariosto) | 5.30pm -7.00pm |
Topics |
Slides and additional material | |
| Introduction | Fundamental Concepts of a Distributed Systems: hardware, softwareand communication technologies. From system specification to system implementation | slides, T1 |
| From Concurrent to Distributed Systems. | paper1, paper2 , (lecture notes in italian) | |
| Basic Abstractions
|
processes: events, histories, failures, distributed computations, distributed algorithms, safety and liveness properties. | T1 |
| system models : Sinchronous, Asynchronous, Eventually synchronous | T1 | |
| point-to-point communication channels : fair loss, stubborn, perfect. | T1, slides (pdf) | |
| Failure detectors: perfect failure detectors, eventually perfect failure detectors, leader elections with failure detectors. | T1 | |
| Distributed Computations | happened-before relation, vector and logical clocks. Application of Logical Time: Causal Communication, message stability | T1, "Fundamentals of Distributed Computing: A practical tour of vector clocks" |
| Synchronization |
Register: non-atomic operations, specification (safe, regular, atomic), basic protocols | T1, slides (1, 2, 3) |
| Consensus: definition, FLP impossibility result, non-uniform flooding consensus, uniform flooding consensus, rotating coordinator protocol. Application of consensus: Total order protocol, non-blocking atomic commitment. | T1, slides (pdf) | |
| Communication |
Specification and Protocols: non-uniform (regular) Reliable Broadcast, uniform reliable broadcast, Causal Broadcast, Total Order Broadcast, hierarchy of total order broadcast. Probabilistic broadcast. | |
| Replication | Software Replication : Primary backup, active replication. Virtual synchrony. | |
| paper: "Fault Tolerance by Replication in distributed systems" (pdf) | ||
| Information Diffusion | Sistemi Publish/Subscribe. Il paradigma di comunicazione, event routing, SIENA, SCRIBE, TERA. | |