Master Database Sharding: Solving Read Diffusion with Range and Modulo Strategies

This article explains why large MySQL tables degrade performance, introduces vertical and horizontal sharding techniques—including range‑based and modulo‑based partitioning—covers the read‑diffusion problem caused by non‑sharding keys, and presents practical solutions such as auxiliary index tables, Elasticsearch integration, and TiDB as a distributed alternative.

ElasticsearchRange ShardingTiDB

0 likes · 15 min read

Master Database Sharding: Solving Read Diffusion with Range and Modulo Strategies

vivo Internet Technology

Feb 26, 2025 · Backend Development

Building a Million-User Group Chat System: Server-Side Architecture and Implementation

The article details how to engineer a Web‑based group chat that supports one million members by selecting WebSocket for real‑time communication, using read‑diffusion storage, a three‑layer architecture with Redis routing and Kafka queues, ensuring ordered, reliable delivery via TCP, ACKs and UUID deduplication, calculating unread counts with Redis ZSETs, and handling massive traffic through rate‑limiting, protobuf compression and message chunking.

IM SystemMessage OrderingWebSocket

0 likes · 21 min read

Building a Million-User Group Chat System: Server-Side Architecture and Implementation

Baidu Tech Salon

May 10, 2023 · Backend Development

Full Volume Message Push Implementation Based on Public Mailbox at Baidu

Baidu’s full‑volume push system leverages a public‑mailbox architecture that merges a single shared mailbox with users’ private mailboxes during retrieval, enabling low‑disturbance, high‑speed delivery to over 600 million monthly active users while reducing write costs, as demonstrated by three pandemic‑policy messages reaching more than 200 million users within three days.

Baidu appInstant MessagingMessage Push

0 likes · 13 min read

Full Volume Message Push Implementation Based on Public Mailbox at Baidu

Tencent Cloud Developer

Aug 10, 2021 · Backend Development

Backend Storage Design for Timeline Feed Streams: Read Diffusion, Write Diffusion, and Hybrid Approaches

Backend storage for timeline feeds can use read‑diffusion (pull), write‑diffusion (push), or a hybrid mix, each with trade‑offs in read/write load, scalability, and storage, while pagination should rely on last_id and snapshots for state‑driven ordering such as live‑streaming feeds.

live streamingpaginationread diffusion

0 likes · 17 min read