Scaling

Tzylo Auth CE is designed to be horizontally scalable when deployed with the correct supporting infrastructure.

Stateless design

The Auth CE service itself is stateless.

• No session data is stored in memory
• Multiple instances can run in parallel
• Requests can be routed to any instance

Shared dependencies

All Auth CE instances must share the same backing services:

• A single database (or database cluster)
• A single Redis instance or cluster

Load balancing

A load balancer should be placed in front of Auth CE instances.

• Round-robin or least-connections strategies work well
• Sticky sessions are not required
• Health checks should be enabled

Cookie and CORS considerations

When scaling across domains or subdomains:

• Configure COOKIE_SAME_SITE correctly
• Use HTTPS in all environments
• Set explicit CORS_ORIGIN values

Database scaling

Database performance is critical for authentication systems.

• Use connection pooling
• Monitor slow queries
• Enable read replicas if needed

Redis scaling

Redis is central to scalable Auth CE deployments.

• Use a managed Redis service when possible
• Monitor memory usage
• Configure eviction policies carefully

Failure handling

Plan for partial failures:

• Restart failed instances automatically
• Alert on database or Redis outages
• Gracefully handle rate limit or auth failures