Course contentsShow
← Cloud and Infrastructure
Lesson 2478 of 2,98953. High Availability PatternsFree lesson

Multi-AZ Deployments

Deploying databases across availability zones for infrastructure-level fault tolerance and automatic failover.

Multi-AZ Deployments

What you'll learn: How to deploy databases across multiple availability zones to achieve infrastructure-level fault tolerance with automatic failover.

The Core Concept

Multi-AZ deployments distribute your database infrastructure across multiple isolated data centers (availability zones) within the same region. When you deploy your primary database in one AZ and maintain a synchronous standby replica in another AZ, you protect against entire data center failures—power outages, network disruptions, or natural disasters affecting a single location.

Think of it like having two identical restaurant kitchens in different buildings connected by a high-speed communication line. If one kitchen floods, the other instantly takes over service without customers noticing.

How It Works

In a Multi-AZ setup:

  • Your primary database handles all read/write operations in AZ-A
  • A synchronous replica runs in AZ-B, receiving every committed transaction
  • Cloud providers maintain a single DNS endpoint that automatically redirects to the healthy instance
  • When AZ-A fails, the system performs automatic failover to AZ-B (typically 60-120 seconds)
  • No data loss occurs because replication is synchronous

This differs from primary-replica setups where you might use asynchronous replication across regions. Multi-AZ prioritizes zero RPO (no data loss) over geographic distribution, accepting slightly higher write latency due to synchronous commits.

When to Use Multi-AZ

Deploy Multi-AZ when:

  • Production workloads require high availability
  • Your RTO allows brief failover windows (1-2 minutes)
  • You cannot tolerate data loss (RPO = 0)
  • Infrastructure failures must be transparent to applications

Key Takeaway: Multi-AZ deployments provide infrastructure-level resilience by automatically failing over to a standby replica in a different availability zone, ensuring zero data loss and minimal downtime during data center failures.