What is MongoDB's architecture made of?

MongoDB's architecture consists of the following key components:

#1. MongoDB Server (mongod)

The main server process. Responsible for storing data, handling queries, replication, sharding, and security. Inside it live several subsystems:

• Storage Engine, the storage module (WiredTiger by default)
• Query Processor, query handling and optimization
• Replication Manager, manages replication
• Sharding Router/Config, the logic for distributing data across shards
• Journal/Write-Ahead Log, the log used for fault tolerance

#2. mongos (the router in a sharded cluster)

A proxy process that routes the application's requests to the right shards, hiding the fact that data is distributed.

#3. Storage Engine (e.g. WiredTiger)

Responsible for the physical storage of data on disk, compression, cache management, and working with the journal and locks.

#4. Replica Set

A group of servers with one primary (writes go to it) and several secondaries (hold copies and can serve reads). Provides fault tolerance and automatic failover.

#5. Sharded Cluster

Horizontal scaling. Made up of:

• Shards, sets of replicas each holding part of the data
• mongos, the query routers
• Config servers, holding metadata about how the data is distributed

#6. Optional integrations

• Client drivers (Node, Python, Go, and others), for applications to talk to the server
• Ops Manager / Atlas, monitoring, management, automation
• Security components, roles, authentication, TLS, auditing

Altogether, MongoDB's architecture is a combination of mongod processes, mongos and config servers when needed, plus the storage and replication subsystems that provide scalability, fault tolerance, and availability.