To serve billions of users with real‑time timelines, modern social platforms combine push‑based delivery for regular users and pull‑based retrieval for high‑profile accounts, employing hot‑cold separation, Kafka fan‑out, Redis caching, and scalable storage strategies to balance write and read loads.
This guide explains how to configure Docker and Containerd to correctly set image push and pull parameters for both HTTP and HTTPS registries, covering insecure registry settings, certificate management, host resolution, and command‑line examples using docker, ctr, crictl, and nerdctl to ensure efficient and secure container image handling.
This article analyzes feed‑timeline architectures, compares pull‑based and push‑based models, proposes an online‑push/offline‑pull hybrid, and details practical implementations using Redis SortedSets, multi‑level caching with Cassandra, cursor‑based pagination, and large‑scale push task sharding.
This article examines the challenges of live streaming push‑pull technology and presents a high‑availability system architecture, network and CDN optimizations, disaster‑recovery mechanisms, fault‑perception monitoring, and security measures to ensure low latency, reliability, and scalability for large‑scale live events.
This article explains the concepts, classifications, design choices, and Java‑Redis implementation of a scalable feed stream system for enterprises, covering initialization, push/pull update mechanisms, handling of new posts, deletions, and follow/unfollow actions.
This article details the evolution of Bilibili's dynamic feed system, outlining its business characteristics, multi-layer microservice architecture, caching strategies, and the transition from incremental pull to a hybrid push‑pull model to address scalability, latency, and data consistency challenges.
This article examines the design of a Weibo‑style feed system, detailing storage tables, the trade‑offs between push (write‑fan‑out) and pull (read‑fan‑out) delivery models, their scalability challenges, and hybrid solutions using caching to achieve low latency and manageable storage costs.
This article breaks down the functional requirements, data models, layered architecture, push‑pull messaging choices, and message flow for building a robust one‑to‑one instant‑messaging service using Redis, MySQL and asynchronous queues.
RocketMQ consumers initialize by validating configuration, creating subscription data, establishing a client instance, loading offsets, and starting services before registering with the client factory, after which they can receive messages via push (default low‑latency) or pull modes, support ordered processing, and handle acknowledgments with configurable retries and dead‑letter handling.
This article explains how to design a service registry that enables service registration, discovery, high availability, and dynamic handling of service instances in a microservice architecture, covering registration methods, consumer/provider interaction, push/pull mechanisms, long‑polling, and heartbeat health checks.
This article provides a comprehensive tutorial on installing ZeroMQ, explains its basic socket types and messaging patterns such as PUB‑SUB, REQ‑REP, and PUSH‑PULL, and offers practical notes, code snippets, and configuration tips for effective use in backend applications.
This article examines how to design a reliable group messaging system with read‑receipt support, comparing push and pull delivery, detailing data structures, acknowledgment flows, and optimization techniques to reduce traffic and storage while maintaining real‑time user experience.
This article explores the core concepts behind Twitter’s feed architecture, explaining how feeds and timelines are built from multiple lists, comparing push and pull notification models, discussing scalability constraints, and offering practical guidelines for choosing the right approach in large‑scale distributed systems.
This article examines the push and pull models used by micro‑blogging platforms such as Twitter and Sina Weibo, analyzes their scalability challenges, and proposes a time‑partitioned pull approach that reduces database load while maintaining fast feed retrieval for active users.
