Prologue to distributed computing

Introduction to cloud computing l.jpg
1 / 39
0
0
1366 days ago, 621 views
PowerPoint PPT Presentation
Distributed computing is a style of registering in which progressively versatile and frequently virtualized assets are given as a serve over the Internet. ...

Presentation Transcript

Slide 1

´╗┐Prologue to distributed computing Jiaheng Lu Department of Computer Science Renmin University of China www.jiahenglu.net

Slide 2

Cloud figuring

Slide 4

Review: What is distributed computing? Distributed computing is a style of registering in which powerfully adaptable and regularly virtualized assets are given as a serve over the Internet. Clients require not know about, ability in, or control over the innovation foundation in the "cloud" that backings them.

Slide 5

Review: Characteristics of distributed computing Virtual. programming, databases, Web servers, working frameworks, stockpiling and systems administration as virtual servers. On request. include and subtract processors, memory, arrange transfer speed, stockpiling.

Slide 6

Review: Types of cloud administration SaaS Software as a Service PaaS Platform as a Service IaaS Infrastructure as a Service

Slide 8

Any question and any remarks ? 2014/3/10 8

Slide 9

Google App Engine

Slide 10

Review: Google App Engine Does one thing great: running web applications Simple application setup Scalable Secure 10

Slide 11

Any question and any remarks ? 2014/3/10 11

Slide 12

Distributed framework

Slide 13

A disseminated framework B A C D

Slide 14

customer server framework Client Server Client

Slide 15

various servers Server

Slide 16

Why conveyed frameworks? What are the favorable circumstances? circulated vs centralized? multi-server vs customer server?

Slide 17

Why dispersed frameworks? What are the focal points? conveyed vs centralized? multi-server vs customer server? Geology Concurrency => Speed High-accessibility (if disappointments happen).

Slide 18

Why not dispersed frameworks? What are the hindrances? circulated vs centralized? multi-server vs customer server?

Slide 19

Why not appropriated frameworks? What are the burdens? conveyed vs centralized? multi-server vs customer server? Costly (to have excess) Concurrency => Interleaving => Bugs Failures prompt to inaccuracy.

Slide 20

Outline Concepts and Terminology What is Distributed information & objects Distributed execution Three level designs Transaction ideas

Slide 21

What's a Distributed System? Concentrated: everything in one place remain solitary PC or Mainframe Distributed: a few sections remote dispersed clients circulated execution disseminated information

Slide 22

Why Distribute? No best association Companies continually swing between Centralized: center, control, economy Decentralized: versatile, responsive, focused Why convey? reflect association or application structure enable clients/makers enhance benefit (reaction/accessibility) disseminated stack utilize PC innovation (financial aspects)

Slide 23

What Should Be Distributed? Clients and User Interface Thin customer Processing Trim customer Data Fat customer Will talk about tradeoffs later Presentation work process Business Objects Database

Slide 24

Transparency in Distributed Systems Make disseminated framework as simple to utilize and oversee as a brought together framework Give a Single-System Image Location straightforwardness: conceal reality that question is remote shroud certainty that protest has moved conceal truth that question is parceled or duplicated Name doesn't change if question is repeated, apportioned or moved.

Slide 25

Outline Concepts and Terminology Why Distribute Distributed information & objects Partitioned Replicated Distributed execution remote strategy call lines Three level designs Transaction ideas

Slide 26

strings shared memory Distributed Execution Threads and Messages Thread is Execution unit (programming simple of cpu+memory) Threads execute at a hub Threads impart by means of Shared memory (nearby) Messages (neighborhood and remote) messages

Slide 27

Peer-to-Peer or Client-Server Peer-to-Peer is symmetric: Either side can send Client-server customer sends demands server sends reactions basic subset of distributed demand reaction

Slide 28

y = pObj->f(x); x f() return val; y = val; val Remote Procedure Call: The way to straightforwardness Object might be neighborhood or remote Methods on protest work wherever it is. Nearby conjuring

Slide 29

intermediary Obj Local? x stub marshal un marshal x un marshal pObj->f(x) marshal Obj Local? x f() return val; val Remote Procedure Call: The way to straightforwardness Remote conjuring y = pObj->f(x); f() return val; y = val;

Slide 30

Transaction Object Request Broker (ORB) Registers Servers Manages pools of servers Connects customers to servers Does Naming, ask for level approval, Provides exchange coordination (new component) Old names: Transaction Processing Monitor, Web server, NetWare Object-Request Broker

Slide 31

Outline Concepts and Terminology Why Distributed information & objects Distributed execution remote methodology call lines Three level designs what why Transaction ideas

Slide 32

Fat Thin Fat Thin Work Distribution Spectrum Presentation and modules Workflow oversees session & summons objects Business objects Database Presentation work process Business Objects Database

Slide 33

PC Evolution to Three Tier Intelligence moved to server Stand-alone PC (incorporated) PC + File & print server message per I/O PC + Database server message per SQL proclamation PC + App server message per exchange ActiveX Client, ORB ActiveX server, Xscript IO ask for answer circle I/O SQL Statement Transaction

Slide 34

The Pattern: Three Tier Computing Presentation Clients do presentation, assemble input Clients do some work process (Xscript) Clients send abnormal state solicitations to ORB (Object Request Broker) ORB dispatches work processes and business objects - intermediaries for customer, arrange streams & lines Server-side work process scripts approach circulated business items to execute errand work process Business Objects Database

Slide 35

Web Client HTML VB Java modules VBscritpt JavaScrpt Middleware ORB TP Monitor Web Server... Protest server Pool VB or Java Script Engine VB or Java Virt Machine HTTP+ DCOM ORB Internet DCOM (oleDB, ODBC,...) LU6.2 Legacy Gateways IBM The Three Tiers Object & Data server .

Slide 36

Why Did Everyone Go To Three-Tier? Presentation Manageability Business rules must be with information Middleware operations apparatuses Performance (adaptability) Server assets are valuable ORB dispatches solicitations to server pools Technology & Physics Put UI handling close client Put shared information preparing close shared information work process Business Objects Database

Slide 37

What Middleware Does ORB, TP Monitor, Workflow Mgr, Web Server Registers exchange programs work process and business objects (DLLs) Pre-allots server pools Provides server execution environment Dynamically checks power (ask for level security) Does parameter restricting Dispatches solicitations to servers parameter restricting burden adjusting Provides Queues Operator interface

Slide 38

Server Side Objects Easy Server-Side Execution A Server ORB gives straightforward execution environment Object gets begin summon shutdown Everything else is programmed Drag & Drop Business Objects Network Receiver Queue Management Connections Security Context Configuration Thread Pool Service rationale Synchronization Shared Data

Slide 39

Q&A Thanks

SPONSORS