Prepare for system design interviews with this guide to designing a WhatsApp-like messaging app. Learn about architectural considerations, API design, scalability strategies, and potential bottlenecks.

Game Central is a platform offering  insights, reviews, and news updates on the gaming industry. From new releases and gaming hardware reviews to interviews with industry experts, GC covers everything gaming enthusiasts need to stay informed and entertained. Readers can delve into detailed analyses of game mechanics, storytelling techniques, and emerging trends, gaining  insights into the ever-evolving world of gaming.

gitconnected

This post outlines how to design a high-level system architecture for a messaging app similar to WhatsApp. Key points include handling 10 billion messages daily with plans to double, using RESTful APIs for compatibility, employing WebSockets for real-time communication, and using message queues like Kafka or RabbitMQ for efficient message distribution. The databases considered are NoSQL options like Cassandra and DynamoDB, emphasizing sharding and partitioning for scalability. End-to-end encryption, handling peak traffic, and future enhancements like group chats and media handling are also discussed.

System Design Interview: Design WhatsApp

Conclusion and current system bottlenecks

Some conception on WhatsApp is that first they don’t use HTTP requests to send messages but rather use Web sockets as the main method for sending. Also they don’t store any messages on their server. When a user sends a message, if the receiver is online and connected to the web socket they would receive the message immediately. If the user is offline a push notification would be sent to their device and tell it to connect to the socket in the background to receive the message. Until then the message is stored in a cache memory encrypted securely.

This post exactly highlights the problem I have with most System Design articles, Coming from a company which has seen petabytes of scale you rarely employ Rest APIs (unless you are stupid and deploy 400-500 replicas of the same) for ingestion, rather there are tools like MQTT (which Whatsapp actually uses to ingest messages) are efficient technologies to reduce both infra and HTTP overhead.

The real gist is in the comments always

Is nosql a only solution for designing high intensive application can’t SQL do the job?