Understanding Geo-Distributed Active-Active Architecture: Principles, Risks, and Implementation Strategies

01 System Availability

To understand geo‑distributed active‑active, we start with three architectural principles: high performance, high availability, and easy scalability. Availability is measured by MTBF and MTTR, with the formula Availability = MTBF / (MTBF + MTTR) * 100%.

Failures can be hardware, software, or force‑majeure, and rapid recovery is essential.

02 Single‑Machine Architecture

A simple single‑instance deployment is vulnerable to data loss; backup can mitigate loss but introduces recovery time and data staleness.

03 Master‑Slave Replication

Adding a replica provides real‑time synchronization, higher data integrity, fault tolerance, and read‑performance improvement.

04 Uncontrollable Risks

Even with redundancy, risks remain at the rack, switch, and data‑center levels; failures in a single data‑center can still cause outages.

05 Same‑City Disaster Recovery

Deploy a second data‑center in the same city, connect via a dedicated line, and use either cold backup (periodic copy) or hot backup (real‑time replica) to ensure data safety.

06 Same‑City Active‑Active

Both data‑centers serve traffic simultaneously, requiring read‑write separation and careful routing to avoid write conflicts.

07 Two‑City Three‑Center

Introduce a third, geographically distant data‑center for disaster backup, typically using cold backup to protect against city‑level catastrophes.

08 Pseudo Cross‑City Active‑Active

Simply mirroring active‑active across cities leads to high latency and performance degradation due to cross‑region data access.

09 True Cross‑City Active‑Active

Each data‑center must host its own primary databases and synchronize data bidirectionally using middleware (e.g., Canal, RedisShake, MongoShake) to avoid latency and ensure consistency.

10 Implementing Active‑Active

Route users at the edge based on business type, hash partitioning, or geographic location so that a user’s requests stay within a single data‑center, eliminating cross‑region conflicts.

11 Geo‑Distributed Multi‑Active

Scale the active‑active model to multiple regions using a star topology with a central hub for data synchronization, achieving high availability, scalability, and rapid failover.

Summary

The article emphasizes high performance, high availability, and easy scalability as core architectural goals, explains redundancy techniques from backup to multi‑region active‑active, and provides practical guidance for building resilient, globally distributed systems.