Transcript Document
Distributed Computing Systems 電機三 B90901047 陳家宏 Outlines • • • • Introduction to Distributed System Middleware layers Application -- Web Services OS support Distributed Systems :Why? • Parallel applications -scalability • Resource sharing -disks, CPU, memory -services ( e.g., name servers, games, etc.) -data (e.g., file systems, databases, etc.) Challenges • Heterogeneity – Networks, OS, hardware, language • Openness – Documents and specifications • Security – Encryption, DOS( denial of service ) • Scalability – Data should be structured hierarchically Challenges (con ‘t) • Failure handling – Any process, computer fails indep • Concurrency – Concurrent requests to resources • Transparency – the system is perceived as a whole Architectural model Applications, services Middleware Operating System Computer and network hardware Middleware layers RMI and RPC Request-reply protocol marshalling and external data representation Middleware layers (con’ t) • Communicate using OS calls -using send() and receive • Communicate using procedure -RPC : remote procedure calls system translates RPC to send/receive • Communicate using object -RMI : remote method invocations Naming of remote data Distributed object Model C A E B F D Middleware layers (con’ t) • Traditional distributed systems • CORBA –common object request broker architecture - cross language • Java RMI - limits to java Applications • Create a web service by yourself! A Web Service Client Listener SOAP (web server) Web Service Object Object Web Services • • • • • New generation distributed systems Built on open technologies and protocols Scalable, can build systems incrementally Less complicated and easier to program Future – Web services are becoming fundamental building blocks in the move to distributed computing Web Services (con’ t) • UDDI – Discovery mechanism • WSDL – Description language -> defines the web service interface • SOAP – Communication protocols • XML & HTTP – data representation format SOAP = 肥皂？ OS support • Alternative server threading architecture • Microkernels & monolithic kernels Thread concept Activation Stack Process Heap Text ( program code) System-provided resources ( socket , files) Alternative server threading architecture Worker thread Remote object Thread per connection I/O thread Thread per object Monolithic kernel S4 Server program S1 S2 S3 Kernel can contain some server processes that execute within its address space, including file servers and some networking Microkernel Dynamic loaded server program S1 S2 Kernel S3 Middleware Language support subsystem OS emulation subsystem Microkernel Hardware … References • Distributed systems concepts and design – George Coulouris, Jean Dollimore, Tim Kindberg