Distributed Systems Explained | System Design Interview Basics

Summary

This video explains the concept of distributed systems, highlighting their crucial role in handling millions of users and petabytes of data in today's online services. It contrasts distributed systems with centralized systems, showcasing how distributed systems can scale horizontally to avoid single points of failure and manage high demand. The script discusses the need for these systems to maintain shared state and collaborate across machines through networking. By decoupling processes into different machines, these systems ensure high availability and low latency, making them essential for large-scale applications and services.

Takeaways

• 😀 Distributed systems are everywhere, hidden behind online services like shopping websites and apps, handling millions of users and petabytes of data seamlessly.
• 😀 A well-designed distributed system makes the user experience feel like interacting with a single system, despite being spread across many machines.
• 😀 Cloud hosting platforms are themselves distributed systems, designed to handle scaling needs for companies and software developers.
• 😀 A centralized system is the opposite of a distributed system, and analyzing centralized systems helps highlight the advantages of distributed ones.
• 😀 Vertical scaling (upgrading a single machine with more CPU and RAM) is limited, and often fails to handle increased load efficiently, especially with growing user bases.
• 😀 Vertical scaling can delay the issue of overload but doesn’t eliminate the single point of failure, and often leads to high latency for users in distant regions.
• 😀 A distributed system scales horizontally, meaning it can grow and shrink on demand, addressing single points of failure and latency issues.
• 😀 In distributed systems, processors (computers) communicate via networks, sharing state and working together toward common goals across different machines.
• 😀 A process is an instance of a running application, which is isolated and separate from other processes, and not part of a distributed system if confined to a single machine.
• 😀 The goal in designing distributed systems is to ensure processes can collaborate effectively, communicating and coordinating through the network, sharing state, and avoiding single points of failure.

Q & A

What is a distributed system?

A distributed system is a system composed of multiple processors running on different computers that communicate and coordinate their actions through a network. These systems work together to achieve a common goal and are designed to scale and maintain high availability.

How do distributed systems benefit users?

Distributed systems provide users with a seamless experience, making it appear as if they are interacting with a single system, even though it might be made up of many different machines working together.

What is the difference between a distributed system and a centralized system?

A centralized system is one where all components and processes are located on a single machine or server, whereas a distributed system spreads components across multiple machines, offering greater scalability, reliability, and fault tolerance.

What is vertical scaling in a centralized system?

Vertical scaling involves upgrading the capacity of a single machine (e.g., adding more CPU, RAM, or storage) to handle increased load. However, it has limitations since there's a finite amount of resources that can be added to one machine.

Why is vertical scaling limited in handling large-scale services?

Vertical scaling is limited because there is only so much power that can be added to a single machine. Once the machine reaches its capacity, it can no longer handle additional load, leading to performance issues or even service outages.

What is horizontal scaling in a distributed system?

Horizontal scaling involves adding more machines to a system to distribute the load and increase capacity. This allows a system to scale infinitely and handle a large number of users or requests without the limitations of a single machine.

What is a single point of failure in a centralized system?

A single point of failure is a situation where a single machine or component is critical to the operation of the entire system. If it fails (e.g., due to power outage), the entire service becomes unavailable.

How do distributed systems address single points of failure?

In a distributed system, the workload is spread across multiple machines. If one machine fails, other machines can take over, ensuring the system remains available and operational, thus eliminating the single point of failure.

Why is the network important in a distributed system?

The network is essential in a distributed system because it enables communication between different machines. Processes on different machines must exchange information and coordinate their actions to ensure the system functions properly.

What does it mean to maintain a shared state in a distributed system?

Maintaining a shared state in a distributed system means that the processes across different machines must have access to consistent, synchronized data to work together toward achieving their common goal. This ensures that the system operates correctly and avoids conflicts or inconsistencies.