Chapter 1: Introduction s7

CHAPTER 1: INTRODUCTION

CHAPTER OUTLINE

 System Definitions  Hardware Concepts  Software Concepts  Client/Server Model

System Definitions

Traditional Uniprocessor Operating Systems

 Resource Manager (system’s view) – processor, information, memory, device  Extended Machine (user’s view) – virtual concepts

With goals: efficiency, flexibility, robustness

Recent Technological Advances

 Widespread low-cost high-performance PCs  High-speed network interconnection

Emerging Applications

 Concurrent/parallel  Distributed  Collaborative

Distributed Systems to Support New Applications

 (user’s view) – Transparency: a single computer view of multiple computer systems  (system’s view) – Distributions: decentralized/autonomous, cooperative/collaborative Unique Problems (Limitations) in Distributed Systems

 lack of global state information  lack of global clock (or time)  failures of system components (processors, communication networks)

Transparency - access, location, migration, relocation, replication, concurrency, failure, persistence

Interoperability – interface, standard, openness, separation of model/policy/mechanism

Middleware – organization for implementation of distributed systems

 Models – file, communication (RPC), object  Layers – system, application

Underlying Operating Systems for Distributed Systems

 Distributed Operating Systems – transparency for multiprocessor and homogeneous multiple computer systems  Network Operating Systems – interoperability/portability/openness for heterogeneous systems and networks

Additional goals:

 Scalability – size, geographical distance, administrative complexity – asynchronous communication, distribution, caching  Robustness – security, reliability, privacy – cryptographic schemes, fault-tolerance, dispersion Hardware Concepts

Shared Memory Multiprocessor Systems

 Bus – tightly coupled  Switched – cross-point, interconnection network (e.g., Omega network), etc.

Homogeneous Multiple Computer Systems

 Bus – tightly/loosely coupled (through message passing)  Switched – grid, cube, interconnection network – routing

Emulating Shared-Memory in Message-Passing Systems (DSM)

 Migration  Replication  Thrashing and false sharing

Message Passing Heterogeneous Multiple Computer Systems

 Networks

Software Concepts

Major Issues in Uniprocessor OS

 Virtual machine  Multi-level protection (mode)  Minimum kernel (microkernel)

Additional Issues in Multiprocessor DOS

 Multiprocessor scheduling  Synchronization – semaphore, monitor, and etc. Additional Issues in Multiple Computer DOS

 Communication primitives – send/receive – naming, buffering and blocking  Distributed synchronization  Relationship between synchronization and communication

Major Issues in Network OS (NOS)

 Network protocols for interoperability  Network services – remote access and resource sharing – telnet, ftp, smtp, http, nfs

More Issues for Distributed Processing in Network Environment

 Remote execution and process migration  Load sharing and balancing  Collaborative/cooperative application

 MIDDLEWARE-BASED DISTRIBUTED SYSTEMS

Middleware: scalability and openness of NOS and transparency of DOS

 Services: locating services, group management, authentication/authorization, etc.  Communication: reliable multicast, atomic transaction, etc.

Middleware Models

 NFS  RPC  Distributed Objects  Distributed Documents Client/Server Model

 Interaction > communication > access resource > request/reply  Dumb/thin/heavy weight client  Vertical distribution: user interface/application server/database server  Horizontal distribution: parallelizing servers  Parallelizing clients: P2P

Some Emerging Computing Paradigms

 Ubiquitous/pervasive computing  Mobile agent-based computing  Mobile/wireless computing  Grid computing  P2P computing