Overview

Introduction

Passwork supports a fault-tolerant architecture that provides high availability and keeps the service running even if individual infrastructure components fail.

Fault tolerance allows you to:

• Ensure continuous service for users
• Distribute load across multiple servers
• Scale the system to handle more users and higher throughput
• Reduce operational risks by keeping behavior predictable during failover and incidents

warning

Fault tolerance in Passwork is available only with the advanced license.

Principles of fault tolerance

A fault-tolerant Passwork deployment is built on two core principles:

1. Redundant application servers — several application servers run in parallel, so the service stays available even if one or more nodes fail.
2. Database replication — a MongoDB replica set keeps data synchronized across nodes and elects a new Primary automatically.

Minimum requirements:

• At least 2 application servers behind a load balancer
• At least 3 MongoDB nodes in a replica set
• Independent infrastructure for database nodes (separate racks/data centers when possible)
• Configured health checks and connection monitoring

General architecture of a fault-tolerant system:

Scalability

Passwork supports two scaling models to increase performance and handle more users.

Horizontal scaling

Horizontal scaling (scale-out) means adding new servers to the pool to spread the load.

Advantages:

• Higher throughput by distributing requests across servers
• High availability — if one server fails, others keep working
• Gradual, incremental expansion

Use cases:

• Adding new Passwork application servers
• Expanding the MongoDB replica set with additional secondary nodes

Vertical scaling

Vertical scaling (scale-up) means increasing resources (RAM, CPU, storage) on an existing server.

Advantages:

• Quick performance boost without changing the architecture
• Simple to implement — no balancing or synchronization required
• Suitable for small and mid-size loads

Limitations:

• Hardware limits (max RAM/CPU on a single host)
• Single point of failure — if the server goes down, the system is unavailable
• A powerful single server can cost more than several smaller ones

Recommendations

• For fault-tolerant deployments, prefer horizontal scaling
• Vertical scaling fits early-stage setups or environments without strict HA requirements
• A combined approach also works: scale servers up when needed and scale out for availability

Components of the fault-tolerant architecture

A fault-tolerant Passwork deployment consists of two main components.

1. Multiple application servers

Using several Passwork application servers behind a load balancer provides:

• Automatic traffic redistribution if a server fails
• Even load distribution across healthy servers
• Horizontal scalability

Learn more about architecture, load balancing, and health checks in application servers.

2. Database replication

Using a MongoDB replica set provides:

• Automatic elections and Primary replacement when the current Primary fails
• Data synchronization between nodes
• High data availability

More details are in database.

Benefits of a fault-tolerant architecture

High availability

The system keeps running even when individual components fail:

• If an application server goes down, the load balancer routes traffic to the remaining servers
• If the Primary database node fails, the replica set elects a new Primary, and application servers start using it

Scalability

A fault-tolerant architecture makes it easy to scale:

• Horizontal scaling — add application servers and MongoDB nodes
• Vertical scaling — add resources to existing servers

Performance

Distributing load across multiple servers allows you to:

• Serve more concurrent requests
• Reduce pressure on each server
• Improve response times

LLM version