Top
Print

Distributed Programming in Mozart - A Tutorial Introduction

Peter Van Roy, Seif Haridi, Per Brand and Raphael Collet

This tutorial shows how to write efficient and robust distributed applications with the Mozart programming system. We first present and motivate the distribution model and the basic primitives needed for building distributed applications. We then progressively introduce examples of distributed applications to illustrate servers, agents, mobility, collaborative tools, fault tolerance, and security.

The tutorial is suitable for Oz programmers who want to be able to quickly start writing distributed programs. The document is deliberately informal and thus complements the other Oz tutorials and the research papers on distribution in Oz.

The Mozart programming system has been developed by researchers from DFKI (the German Research Center for Artificial Intelligence), SICS (the Swedish Institute of Computer Science), the University of the Saarland, UCL (the Universit� catholique de Louvain), and others.

Warning: THE MATERIAL IN THIS DOCUMENT IS INCOMPLETE. THIS DOCUMENT WILL EVENTUALLY BE SUPERSEDED BY A COMPLETE DOCUMENT. IN THE MEANTIME, WE RECOMMEND LOOKING AT THE DISTRIBUTED PROGRAMMING (DP) CATEGORY OF THE MOGUL ARCHIVE AND READING CHAPTER 11 OF THE BOOK 'CONCEPTS, TECHNIQUES, AND MODELS OF COMPUTER PROGRAMMING'.

• 1 Introduction

• 2 Distribution Model
  • 2.1 Language entities
    • 2.1.1 Objects
    • 2.1.2 Other stateful entities
    • 2.1.3 Single-assignment entities
    • 2.1.4 Stateless entities
    • 2.1.5 Sited entities
  • 2.2 Sites
    • 2.2.1 Controlled system shutdown
    • 2.2.2 Distributed memory management
  • 2.3 Bringing it all together

• 3 Basic Operations and Examples
  • 3.1 Global naming
    • 3.1.1 Connecting applications by means of tickets
    • 3.1.2 Persistent data structures by means of pickles
    • 3.1.3 Remote computations and functors
  • 3.2 Servers
    • 3.2.1 The hello server
    • 3.2.2 The hello server with stationary objects
    • 3.2.3 Making stationary objects
    • 3.2.4 Making stationary objects, the new way
    • 3.2.5 A compute server
    • 3.2.6 A compute server with functors
    • 3.2.7 A dynamically-extensible server
  • 3.3 Practical tips
    • 3.3.1 Timing and memory problems
    • 3.3.2 Avoiding sending useless data
    • 3.3.3 Avoiding sending classes

• 4 Failure Model
  • 4.1 Fault states
    • 4.1.1 Temporary faults
    • 4.1.2 Permanent faults
    • 4.1.3 Remote problems
  • 4.2 Basic model
    • 4.2.1 The fault stream
      • Stream finalization
      • Variables
  • 4.3 Advanced features
    • 4.3.1 Failure propagation
  • 4.4 Fault states for language entities
    • 4.4.1 Stationary state protocols: stationary, sited
    • 4.4.2 Migratory state protocols: migratory, pilgrim
    • 4.4.3 Replicated state protocol: replicated
    • 4.4.4 Immediate copy protocol: immediate
    • 4.4.5 Non-immediate copy protocol: eager, lazy
    • 4.4.6 Variable protocols: variable, reply

• 5 Fault-Tolerant Examples
  • 5.1 A fault-tolerant broadcast channel
    • 5.1.1 Sample use (no fault tolerance)
    • 5.1.2 Definition (no fault tolerance)
      • Client definition
      • Server definition
    • 5.1.3 Sample use (with fault tolerance)
    • 5.1.4 Definition (with fault tolerance)
      • Client definition
      • Server definition

• 6 Limitations and Modifications
  • 6.1 Performance limitations
  • 6.2 Functionality limitations
  • 6.3 Modifications

• Bibliography

Index

A

annotation: 3.2.4 Making stationary objects, the new way

application

connected: 1 Introduction

overall structure: 2.3 Bringing it all together

asynchronous many-to-one channel: 2.1.2 Other stateful entities

atom: 2.1.4 Stateless entities

B

base module: 2.1.5 Sited entities

basic failure model: 4.2 Basic model

C

cached object: 2.1.1 Objects

cell

analogy to C and Java: 2.1.2 Other stateful entities

centralized semantics: see language semantics (2 Distribution Model)

channel

port (asynchronous many-to-one): 2.1.2 Other stateful entities

chunk: 2.1.4 Stateless entities

C language

relation to Oz cell: 2.1.2 Other stateful entities

class: 2.1.4 Stateless entities

communication failure: 4 Failure Model

component-based programming: 2.3 Bringing it all together

computed functor: 3.1.3 Remote computations and functors

compute server: 3.2.5 A compute server

concurrency

90% rule: 2.1.3 Single-assignment entities

connected applications: 1 Introduction

Connection module: 1 Introduction

Connection module

example: 3.1.1 Connecting applications by means of tickets

constrained variable: 2.1.5 Sited entities

CORBA: 1 Introduction

cross-site cycle: 2.2.2 Distributed memory management

cycle: 2.2.2 Distributed memory management

D

data-flow: 2.1.2 Other stateful entities

distributed semantics

atom: 2.1.4 Stateless entities

chunk: 2.1.4 Stateless entities

class: 2.1.4 Stateless entities

definition: 2 Distribution Model

function: 2.1.4 Stateless entities

functor: 2.1.4 Stateless entities

list: 2.1.4 Stateless entities

name: 2.1.4 Stateless entities

number: 2.1.4 Stateless entities

object: 2.1.4 Stateless entities

object-record: 2.1.4 Stateless entities

procedure: 2.1.4 Stateless entities

record: 2.1.4 Stateless entities

space: 2.1.5 Sited entities

string: 2.1.4 Stateless entities

thread: 2.1.5 Sited entities

weak dictionary: 2.1.5 Sited entities

DP

annotate: 3.2.4 Making stationary objects, the new way

failure model: 4 Failure Model

getFaultStream: 4.2.1 The fault stream

module: 1 Introduction

E

example

using Connection module: 3.1.1 Connecting applications by means of tickets

using Pickle module: 3.1.2 Persistent data structures by means of pickles

using Remote module: 3.1.3 Remote computations and functors

F

fail-silent assumption: 4 Failure Model

failure

basic model: 4.2 Basic model

fail-silent: 4 Failure Model

partial failure property: 4 Failure Model

permanent network problem: 4 Failure Model

temporary network problem: 4 Failure Model

failure handling

fault stream: 4.2.1 The fault stream

fault state: 4.1 Fault states

fault state

localFail: 4.1 Fault states

ok: 4.1 Fault states

permFail: 4.1 Fault states

tempFail: 4.1 Fault states

fault stream

definition: 4.2.1 The fault stream

DP.getFaultStream: 4.2.1 The fault stream

failure handler: 4.2.1 The fault stream

finalization: Stream finalization

feeling

warm, fuzzy: 2.3 Bringing it all together

finite domain variable: 2.1.5 Sited entities

finite set variable: 2.1.5 Sited entities

free record variable: 2.1.5 Sited entities

function: 2.1.4 Stateless entities

functor: 2.1.4 Stateless entities

functor

computed: 3.1.3 Remote computations and functors

future: 2.1.3 Single-assignment entities, 2.1.5 Sited entities

G

garbage collection

distributed: 2.2.2 Distributed memory management

in a server: 3.2.7 A dynamically-extensible server

geographic distribution: 2 Distribution Model

global name

external: 3.1 Global naming

hostname: 3.1 Global naming

internal: 3.1 Global naming

ticket: 3.1 Global naming

URL: 3.1 Global naming

H

halt

abnormal: see shutdown, abnormal (2.2.1 Controlled system shutdown)

normal: see shutdown, normal (2.2.1 Controlled system shutdown)

home site: 2.1.5 Sited entities

hostname

definition: 3.1 Global naming

I

Internet: 1 Introduction

J

Java language: 1 Introduction

Java language

limitation: 3.2.7 A dynamically-extensible server

relation to Oz cell: 2.1.2 Other stateful entities

L

language entity

module: 2.1.5 Sited entities

sited: 2.1.5 Sited entities

unsited: 2.1.5 Sited entities

language semantics: 2 Distribution Model

latency tolerance: 2.1.3 Single-assignment entities

lazy

class: 2.1.1 Objects

lightweight thread: 2.1.1 Objects

list: 2.1.4 Stateless entities

localFail

fault state: 4.1 Fault states

lock

thread-reentrant: 2.1.2 Other stateful entities

logic variable: 2.1.3 Single-assignment entities

logic variable

constrained: 2.1.5 Sited entities

constrained variable: 2.1.5 Sited entities

finite domain: 2.1.5 Sited entities

free record: 2.1.5 Sited entities

read-only (future): 2.1.3 Single-assignment entities, 2.1.5 Sited entities

M

memory management: 2.2.2 Distributed memory management

mobile object: 2.1.1 Objects

module

base

Cell: 2.1.5 Sited entities

definition: 2.1.5 Sited entities

Float: 2.1.5 Sited entities

Int: 2.1.5 Sited entities

List: 2.1.5 Sited entities

Lock: 2.1.5 Sited entities

Number: 2.1.5 Sited entities

Port: 2.1.5 Sited entities

Procedure: 2.1.5 Sited entities

Record: 2.1.5 Sited entities

definition: 2.1.5 Sited entities

system

Application: 2.1.5 Sited entities

Browser: 2.1.5 Sited entities

Connection: 2.1.5 Sited entities

Connection: 1 Introduction

definition: 2.1.5 Sited entities

DP: 1 Introduction

FD: 2.1.5 Sited entities

Module: 2.1.5 Sited entities

Open: 2.1.5 Sited entities

OS: 2.1.5 Sited entities

Pickle: 2.1.5 Sited entities

Pickle: 1 Introduction

Property: 2.1.5 Sited entities

Remote: 2.1.5 Sited entities

Remote: 1 Introduction

Search: 2.1.5 Sited entities

Tk: 2.1.5 Sited entities

multicast: 2.1.3 Single-assignment entities

multitasking: 2 Distribution Model

N

name: 2.1.4 Stateless entities

network-transparent: 2 Distribution Model

network failure: 4 Failure Model

number: 2.1.4 Stateless entities

O

object

cached: 2.1.1 Objects

mobile: 2.1.1 Objects

object-record: 2.1.4 Stateless entities

sequential asynchronous stationary: 2.1.1 Objects

stationary: 2.1.1 Objects, 3.2.3 Making stationary objects, 3.2.4 Making stationary objects, the new way

ok

fault state: 4.1 Fault states

owner site: 2.1.1 Objects, 2.1.5 Sited entities

P

partial failure property: 4 Failure Model

permFail

fault state: 4.1 Fault states

pickle: 3.1 Global naming

Pickle module: 1 Introduction

Pickle module

example: 3.1.2 Persistent data structures by means of pickles

port: 2.1.2 Other stateful entities

port

Oz vs. Unix: 2.1.2 Other stateful entities

procedure: 2.1.4 Stateless entities

process: 2 Distribution Model

programming

component-based: 2.3 Bringing it all together

proxy site: 2.1.1 Objects

R

record: 2.1.4 Stateless entities

record

module: 2.1.5 Sited entities

object-record: 2.1.4 Stateless entities

remote computation

example: 3.1.3 Remote computations and functors

of a statement (needing resources): 3.1.3 Remote computations and functors

of a statement (not needing resources): 3.2.5 A compute server

Remote module: 1 Introduction

Remote module

example: 3.1.3 Remote computations and functors

resource: 2.1.5 Sited entities

S

semantics

distributed: 2 Distribution Model

language: 2 Distribution Model

server: 3.2 Servers

server

compute server: 3.2.5 A compute server

dynamically-extensible: 3.2.7 A dynamically-extensible server

shutdown

abnormal: 2.2.1 Controlled system shutdown

normal: 2.2.1 Controlled system shutdown

site

definition: 2 Distribution Model

multitasking: 2 Distribution Model

owner: 2.1.1 Objects

proxy: 2.1.1 Objects

sited entity

definition: 2.1.5 Sited entities

resource: 2.1.5 Sited entities

space: 2.1.5 Sited entities

thread: 2.1.5 Sited entities

weak dictionary: 2.1.5 Sited entities

space

sited entity: 2.1.5 Sited entities

stateless entity

pickle: 3.1 Global naming

stationary object

definition: 3.2.3 Making stationary objects

definition: 2.1.1 Objects

example: 3.2.2 The hello server with stationary objects

stream: 2.1.3 Single-assignment entities

string: 2.1.4 Stateless entities

system module: 2.1.5 Sited entities

T

tempFail

fault state: 4.1 Fault states

thread: 2.1.2 Other stateful entities

thread

data availability: 2.1.3 Single-assignment entities

lightweight: 2.1.1 Objects

reentrant lock: 2.1.2 Other stateful entities

sited entity: 2.1.5 Sited entities

thread-reentrant lock: 2.1.2 Other stateful entities

ticket

definition: 3.1 Global naming

example: 3.1.1 Connecting applications by means of tickets

time-out

difference with temporary fault: 4.1.1 Temporary faults

U

Unix port: 2.1.2 Other stateful entities

unsited entity: 2.1.5 Sited entities

URL

definition: 3.1 Global naming

V

variable

final in Java: 2.1.3 Single-assignment entities

logic variable: 2.1.3 Single-assignment entities

W

weak dictionary

sited entity: 2.1.5 Sited entities