Evolution of 58 Bangbang IM System Architecture Across Four Phases
This article chronicles the four-stage evolution of the 58 Bangbang instant messaging platform’s architecture—from a traditional IM system to a merchant management platform, then to a mobile marketing tool, and finally to a high‑throughput mobile push solution—detailing design choices, scaling strategies, and technology stacks used.
The 58 Bangbang platform, part of China’s largest life‑service portal, processes over 40 billion requests daily and supports more than one million concurrent users, prompting continuous architectural evolution to meet growing traffic and product demands.
Stage 1 – Traditional IM: The system comprised a five‑layer backend (Entry, Logic, Router, Data Access, Data Storage) with stateless, horizontally scalable components, handling long‑lived TCP connections for PC and web clients and achieving up to 300 k QPS per node.
Stage 2 – From IM to Merchant Management Platform: New third‑party services (housing, recruitment, etc.) were added via HTTP calls, and the architecture introduced a server‑side transit layer to mediate client requests, reducing client‑side coupling but increasing request latency and operational complexity.
Stage 3 – From Merchant Platform to Mobile Marketing Tool: The platform shifted to lightweight WebService calls for third‑party services, cutting network hops to two per request, improving latency, stability, and troubleshooting while retaining TCP long‑connections for core IM functions.
Stage 4 – Mobile Push System: To address mobile network instability, a push solution combining client polling, SMS, and server‑side long‑connections was evaluated; the final design adopted third‑party push providers together with a custom high‑performance provider, delivering billions of push messages daily with dynamic strategies per device.
Architecture Digest
Focusing on Java backend development, covering application architecture from top-tier internet companies (high availability, high performance, high stability), big data, machine learning, Java architecture, and other popular fields.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.