Distributed Objects in Distributed Systems

Introduction to Distributed Objects

• Distributed Computing: Involves multiple computers or nodes working together to achieve a common task.
• Focus on Distributed Objects: Core components that enable the distribution of methods and data.

Key Concepts in Distributed Systems

• Remote Procedure Calls (RPC): Allows client programs to call procedures on server programs.
• Remote Method Invocation (RMI): Objects invoke methods of objects located on different hosts.
• Event-based Programming Model: Objects receive notifications about events occurring in other objects.

Middleware

• Definition: Software that bridges different applications or layers, enabling them to communicate.
• Functions of Middleware:
  • Facilitates communication between processes and message passing.
  • Operates with multiple protocols and APIs.

Middleware Layers

• Layers involved:
  • Request-Reply Protocol
  • External Data Representation
  • Operating System Interfacing
  • Modules like RMI, RPC, and events.

Benefits of Middleware

• Location Transparency: Remote objects appear as local, simplifying client-server interactions.
• Communication Protocols: Clients do not need to be aware of underlying network protocols (UDP vs TCP).
• Operational Consistency: Differences in data representation across hardware/OS are abstracted away.
• Programming Language Flexibility: Supports interactions between different programming languages for clients and servers.

Interfaces in Distributed Systems

• Defined as modules with strict communication protocols.
• Implementation is not mandatory on the consuming side, promoting modularity.

Types of Interfaces

• Service Interface: List of available procedures for clients.
• Remote Interface: Definition of remote objects and their accessible methods with input/output arguments.

Object Interfaces

• Interfaces denote method signatures (arguments, return values, exceptions) without implementation details.
• They do not have constructors or data.

Interface Definition Languages (IDL)

• Purpose: Specifies parameters, argument types to define remote objects.
• Provides a common specification to facilitate interoperability across programming languages.

CORBA IDL Example

struct Person {
    string name;
    string place;
    long year;
};

interface PersonList {
    readonly attribute string listname;
    void addPerson(in Person p);
    void getPerson(in string name, out Person p);
    long number();
};

Object Model in Distributed Computing

• Objects include state (data) and behavior (methods).
• Often represent real-world entities but can encapsulate more abstract concepts.

Object References

• Objects accessed through references and can be:
  • Assigned to variables.
  • Passed as parameters in method calls.
  • Returned as method results.

Actions in Object-Oriented Programming

• Actions triggered by object methods can:
  • Change the object’s state.
  • Instantiate new objects.
  • Trigger further method invocations.
  • Raise exceptions for error handling.

Exception Handling

• Allows the program to manage unexpected conditions without crashing.
• Uses a try-catch mechanism to handle errors gracefully.

Garbage Collection

• Method for automatically freeing memory of unused objects.
• Implemented in languages like Java, while C++ requires manual handling by the programmer.

Remote Method Invocation (RMI)

• Enables methods of remote objects to be invoked over a network.
• Remote objects must be accessible via a remote reference.

Differences between RMI and General RMIs

• Java RMI is specific to Java; broader RMI is language-agnostic.
• Confusion may arise if using the term RMI generically.

Fault Tolerance in Distributed Systems

• Mechanisms to manage errors within systems, including:
  • Acknowledging errors in client, server, or network.
  • Requires distinct strategies for handling reliability.

Fault Tolerance Strategies

• Retry Request Message: Retransmits requests until a response is clear.
• Duplicate Filtering: Manages multiple identical requests.
• Result Retransmission: May resend results rather than re-executing requests.

Invocation Semantics

• Maybe: Requests may execute once or not at all (common in failures).
• At-least-once: Guarantees execution at least once but risks duplication.
• At-most-once: Ensures execution exactly once without duplicates.

Communication Modules

• Communication Module: Handles message transmission and request processing.
• Remote Reference Module: Translates between remote and local references, sustaining a table of references.

Proxy, Dispatcher, and Skeleton

• Proxy: Local representation of a remote object.
• Dispatcher: Chooses the method for a received request.
• Skeleton: Receives a network call, connecting it to the remote object.

Binder in Distributed Systems

• A service that maps client requests to server object references (e.g., Java's RMI registry).

Events and Notifications in Distributed Systems

• Events: Actions causing changes in object states.
• Notifications: Alerts sent to objects when an event occurs.

Publish-Subscribe Model

• Clients subscribe to servers for event updates.
• Allows selective notifications based on client interests.

Observers

• Intermediate objects that manage event notifications for subscribers:
  • Handle bulk notifications.
  • Filter and prioritize messages.
  • Store notifications until the subscriber is ready.

Example of Event Notification

• The object can publish directly or through an observer, allowing flexible handling of updates.