JSON vs BSON Data Types & MongoDB Internal Storage Engine (WiredTiger)

While developers query MongoDB using familiar JavaScript/JSON notation, the database internally serializes and stores documents as BSON (Binary JSON). Behind the scenes, the WiredTiger storage engine delivers enterprise-grade performance, in-memory caching, and transparent disk compression.

1. Why Not Pure JSON? (JSON vs BSON)

Plain JSON (JavaScript Object Notation) is a lightweight text-based format, but it has severe limitations for a production database:

1. 1
   Limited Data Types: JSON supports only strings, numbers, booleans, arrays, objects, and null. It lacks native representations for Dates, 64-bit integers, regular expressions, and raw binary bytes.
2. 2
   Text Parsing Overhead: Text JSON requires parsing strings character-by-character on every query, which is computationally expensive.
3. 3
   No Traversal Offsets: Finding a field in a large nested JSON document requires scanning the entire string up to that character.

BSON (Binary JSON) solves these issues:

• Binary Encoding: Compact binary representation with prefixed length bytes for fast traversal.
• Rich Type Support: Native ISODate, ObjectId, Decimal128 (high-precision financial arithmetic), Int64, and BinData.

2. Anatomy of the MongoDB ObjectId

Every MongoDB document requires a unique primary key _id. If not provided, MongoDB automatically generates a 12-byte ObjectId:

• Built-in Timestamp: You can extract the exact creation timestamp of any document using doc._id.getTimestamp() without storing an extra createdAt field!
• Natural Ordering: ObjectIds are roughly ordered chronologically.

3. The WiredTiger Storage Engine

MongoDB's default storage engine, WiredTiger, is responsible for managing data between memory and disk:

• Document-Level Concurrency: Provides fine-grained document-level locking for write operations, maximizing multi-core CPU throughput.
• In-Memory LRU Cache: Uses 50% of available server RAM minus 1 GB by default to cache hot indexes and documents.
• Snappy / Zlib Compression: Transparently compresses collection data and indexes on disk, typically saving 50% to 70% of storage costs compared to raw JSON.
• Write-Ahead Logging (Journaling): Commits write operations to an on-disk journal before flushing to data files, guaranteeing durability across sudden server power outages.