Materi 2 - Arsitektur Sistem Terdistribusi

Summary

This lecture discusses distributed systems, focusing on the technologies that keep them running smoothly even with high user demand. The speaker introduces two primary architectures: Master-Slave and Peer-to-Peer, explaining their advantages and challenges. It covers various types of Peer-to-Peer systems, including structured, unstructured, and hybrid models. The lecture also touches on architectural tiers in distributed systems, microservices, and how services are divided into smaller, manageable parts. Furthermore, the importance of layering (application, middleware, and platform) and the communication between them via API and system calls is discussed. The session concludes with a brief overview of each architecture and approach.

Takeaways

• 😀 Distributed systems are designed to ensure high availability and resilience, even when accessed by millions of users at once.
• 😀 The Master-Slave architecture has a central server (master) managing other servers (slaves), but if the master server fails, the whole system crashes.
• 😀 Peer-to-Peer (P2P) architecture is decentralized, where all nodes are equal, making it more resilient but harder to coordinate as more nodes are added.
• 😀 Structured Peer-to-Peer networks use predefined connections between nodes (e.g., Blockchain) to ensure efficient data management.
• 😀 Unstructured Peer-to-Peer networks allow nodes to join freely and connect with others, offering flexibility but more complex coordination (e.g., WhatsApp or Telegram).
• 😀 Hybrid architecture combines Master-Slave and Peer-to-Peer systems, where the initial connection is through a master server, but peer-to-peer communication follows (e.g., Zoom, Meet).
• 😀 In a One-Tier system, everything (interface, logic, database) is on a single server, making it prone to failure if the server crashes.
• 😀 Two-Tier systems separate the interface and logic code from the database, with one tier for each. It can be more scalable depending on device resources.
• 😀 Three-Tier architecture separates the user interface, logic, and database into different servers, optimizing performance and scalability.
• 😀 Microservices split a large application into smaller, independent services (e.g., login, payment) that communicate via APIs, improving scalability and fault tolerance.
• 😀 Vertical layering in distributed systems involves separating the application layer, middleware, and platform layer, with APIs and system calls facilitating communication between layers.

Q & A

What is the key focus of this lecture?

The key focus of the lecture is understanding distributed systems and the architecture behind ensuring their reliability and accessibility even under heavy user load.

What are the two main architectures in distributed systems discussed?

The two main architectures discussed are Master-Slave and Peer-to-Peer.

What is the main advantage of the Master-Slave architecture?

The main advantage of Master-Slave architecture is that data coordination is relatively easier because there is one master server that controls the other servers.

What is a key disadvantage of the Master-Slave architecture?

A major disadvantage of Master-Slave architecture is that if the master server goes down, the entire system can fail, as there is only one master controlling the others.

How does Peer-to-Peer architecture differ from Master-Slave?

In Peer-to-Peer architecture, all nodes (servers) are equal, with no master or slave. Each node can provide the same service, making the system more resilient and scalable, but also more complex in terms of data coordination.

What is the advantage of a Peer-to-Peer system?

The advantage of Peer-to-Peer systems is that they are highly scalable and resistant to failure since all nodes can handle the same tasks, and new nodes can be added or removed easily.

What are the three types of Peer-to-Peer systems mentioned?

The three types of Peer-to-Peer systems mentioned are structured overlay, unstructured overlay, and hybrid.

What is a key characteristic of a structured overlay in Peer-to-Peer systems?

In a structured overlay, new nodes must join a specific, predetermined position in the network, typically using a hashing mechanism to ensure efficiency and fast access to data.

Can you give an example of a system that uses an unstructured overlay?

An example of a system that uses an unstructured overlay is WhatsApp or Telegram groups, where nodes can join any available group without a specific arrangement.

How does a hybrid Peer-to-Peer system work?

In a hybrid Peer-to-Peer system, a master server is initially used for coordination and connection, but once the nodes are connected, they can communicate directly with each other. Examples include Zoom and Google Meet.

What is the concept of 'Microservices' as introduced in the lecture?

Microservices involve breaking down a service into smaller, manageable parts, each responsible for a specific function, like login, payments, or notifications. Each microservice can be handled by separate servers, allowing for easier scalability and maintenance.

What is the purpose of 'layering' in distributed system architecture?

Layering in distributed systems divides the system into distinct levels, such as the application layer, middleware, and platform. This separation allows for more efficient management and better scalability by allowing each layer to focus on specific tasks.