How WE‑CAN Revolutionizes Global Data Transmission with Ultra‑Low Latency and Cost‑Effective Architecture

WE‑CAN Definition and Goals

WE‑CAN (Communications Acceleration Network) is a complex network system built on the public Internet that intelligently schedules resources to improve data transmission quality and reduce cost. Its core goal is to create a universal transmission network that can reliably and efficiently send any data from any point to any other point worldwide without special hardware or dedicated lines.

Key Objectives

Faster than CDN

Cheaper than SD‑WAN

More versatile than RTN

Advantages

Compared with CDN, WE‑CAN not only achieves large‑scale edge distribution but also delivers higher speed. It supports multiple transmission modes beyond RTC, provides high‑reach low‑latency media delivery, and offers optional ARQ, FEC, and other redundancy strategies that are transparent to business.

It enables ultra‑large‑scale video live distribution, reduces bandwidth cost, and approaches CDN cost while achieving RTC‑level real‑time performance.

It also ensures reliable transmission for signaling, IM, and other data, guaranteeing delivery and order.

Its service and protocol feature industry‑leading decoupling and layering, offering a simple global MessageBus interface for distributed messaging.

Performance Highlights

Production data shows:

Daily transmission of billions of messages and hundreds of millions of minutes of media.

Coverage of over 200 regions worldwide with edge nodes in every Chinese province.

In‑network transmission quality exceeding 99.9% and end‑to‑end quality over 99%.

Cross‑regional latency under 250 ms, approaching dedicated line quality.

Graphs below illustrate end‑to‑end quality, jitter, and latency improvements when using WE‑CAN versus direct server connections.

Architecture Overview

The transmission process is divided into three stages: client A to edge server A′, edge server A′ to edge server B′, and edge server B′ to client B. WE‑CAN optimizes two scenarios: Server‑to‑Server (in‑network) and Last‑mile (edge access), each balancing quality and cost.

In‑Network Transmission

Core nodes include Edge (access), Relay (intermediate), and Controller (control). Relays form an approximate full‑mesh network and perform dynamic routing based on real‑time quality probes reported to the Controller. The Controller computes optimal routes and distributes them, enabling intelligent load balancing and rapid response to congestion or failures.

Quality is enhanced through real‑time intelligent routing, ARQ/FEC redundancy, and optional multi‑path transmission. Cost is reduced by leveraging public Internet routing, deploying edge nodes in single‑line data centers, and optimizing bandwidth allocation based on historical peaks.

Edge Access

Edge nodes consist of a cluster of services (Gateway, Broker, Driver, Edge, Monitor, ENS‑Registrar) and are managed by a dashboard platform that schedules resources based on real‑time load and historical data. The scheduler assigns the optimal same‑operator edge node to each request, improving quality while minimizing cost.

Layered Decoupling

WE‑CAN’s protocol stack is layered: Application layer (MessageBus), Transport layer (UDP‑based reliable transmission), and Network layer (routing, congestion control, per‑hop ARQ/FEC). This decoupling allows independent evolution of each layer, simplifies development, and supports flexible service tiers.

Conclusion

By embracing layered decoupling, hierarchical services, and path reuse, WE‑CAN serves as the first transmission foundation independent of business logic, positioning NetEase Cloud Communication as a leader in global intelligent routing networks.