D ISTRIBUTED S YSTEMS [COMP9243] Lecture 2: System Architecture - PowerPoint PPT Presentation

D ISTRIBUTED S YSTEMS [COMP9243] Lecture 2: System Architecture & Communication B UILDING A D ISTRIBUTED S YSTEM Two questions: Slide 1 Slide 3 ➀ Where to place the hardware? ➁ Where to place the software? ➀ System Architectures ➁ Processes & Server Architecture ➂ Communication in a Distributed System ➃ Communication Abstractions System Architecture: ➜ placement of machines ➜ placement of software on machines Where to place?: ➜ processing capacity, load balancing A RCHITECTURE ➜ communication capacity Slide 2 Slide 4 ➜ locality Mapping of services to servers: ➜ Partitioning ➜ Replication ➜ Caching B UILDING A D ISTRIBUTED S YSTEM 1 A RCHITECTURAL P ATTERNS 2

Client-Server from another perspective: Wait for result Client Request Reply A RCHITECTURAL P ATTERNS Slide 5 Slide 7 Server Provide service Time How scalable is this? Example client-server code in C: C LIENT -S ERVER client(void) { struct sockaddr_in cin; Request char buffer[bufsize]; Client Server int sd; Reply Slide 6 Slide 8 sd = socket(AF_INET,SOCK_STREAM,0); connect(sd,(void *)&cin,sizeof(cin)); Kernel Kernel send(sd,buffer,strlen(buffer),0); recv(sd,buffer,bufsize,0); close (sd); } C LIENT -S ERVER 3 C LIENT -S ERVER 4

server(void) { struct sockaddr_in cin, sin; Splitting Functionality: int sd, sd_client; Client machine User interface User interface User interface User interface User interface sd = socket(AF_INET,SOCK_STREAM,0); Application Application Application bind(sd,(struct sockaddr *)&sin,sizeof(sin)); Database listen(sd, queuesize); while (true) { User interface Slide 9 Slide 11 sd_client = accept(sd,(struct sockaddr *)&cin,&addrlen)); Application Application Application recv(sd_client,buffer,sizeof(buffer),0); Database Database Database Database Database DoService(buffer); Server machine send(sd_client,buffer,strlen(buffer),0); (a) (b) (c) (d) (e) close (sd_client); Which is the best approach? } close (sd); } Example client-server code in Erlang: V ERTICAL D ISTRIBUTION (M ULTI - TIER ) % Client code using the increment server client (Server) -> Server ! {self (), 10}, Request Request App. Dbase Client Server Server receive Reply Reply {From, Reply} -> io:format ("Result: ~w~n", [Reply]) Kernel Kernel Kernel end. Slide 10 Slide 12 % Server loop for increment server loop () -> Three ’layers’ of functionality: receive • User interface {From, Msg} -> From ! {self (), Msg + 1}, loop (); • Processing/Application logic stop -> true • Data end. ➜ Logically different components on different machines % Initiate the server start_server() -> spawn (fun () -> loop () end). Leads to Service-Oriented architectures (e.g. microservices). C LIENT -S ERVER 5 V ERTICAL D ISTRIBUTION (M ULTI - TIER ) 6

Vertical Distribution from another perspective: Note: Scaling Up vs Scaling Out? Wait for result User interface (presentation) Horizontal and Vertical Distribution not the same as Horizontal Request and Vertical Scaling . Return operation result Wait for data Application Vertical Scaling: Scaling UP Increasing the resources of a Slide 13 server Slide 15 single machine Request data Return data Horizontal Scaling: Scaling OUT Adding more machines. Database server Time Horizontal and Vertical Distribution are both examples of this. How scalable is this? H ORIZONTAL D ISTRIBUTION P EER TO P EER Front end request Peer handling Replicated Web servers each reply incoming Peer requests containing the same Web pages Kernel Requests Disks handled in Kernel round-robin request reply request fashion reply Peer Slide 14 Slide 16 request Peer reply Kernel Internet Internet Kernel Peer Kernel ➜ Logically equivalent components replicated on different ➜ All processes have client and server roles: servent machines Why is this special? How scalable is this? H ORIZONTAL D ISTRIBUTION 7 P EER TO P EER AND O VERLAY N ETWORKS 8