MongoDB Overview

• Document DB
• supports ACID
• uses JSON/BSON
• distributed scalability (sharding)
• high availability (replica set)
• free open source Community Edition
• pluggable storage engine support
• queries use MQL (MongoDB Query Language)
• has free text search capabilities

Terminology

Compared with Relational databases, MongoDB has the following terms:

BSON

BSON stands for “Binary JSON”, it’s binary-encoded. It sacrifies human-readability for performance. BSON supports more data types than traditional JSON (different numeric types, raw binary, dates, ObjectId, etc.).

Example:

# JSON
{"hello": "world"}

# BSON
\x16\x00\x00\x00           // total document size
\x02                       // 0x02 = type String
hello\x00                  // field name
\x06\x00\x00\x00world\x00  // field value
\x00                       // 0x00 = type EOO ('end of object')

On the “frontend”, we talk with MongoDB in JSON. In the backend, MongoDB stores the documents as BSONs.

All the BSON types can be found in the MongoDB Docs.

Connection

When talking with Mongo DB, it uses TCP/IP with TLS, and the Mongo Wire Protocol. It does not use HTTP. Therefore, when using some tracing or observability in general, you will not see Mongo traffic being recorded by default. You will most likely need some Mongo-specific instrumentation. The Mongo Wire Protocol might be sniffed with tools such as WireShark.

When initiating a connection with Mongo, before any user-initiated query gets sent, the client negotiates with the server on various parameters, which translates to about 4 request-responses (in Mongo Wire Protocol) being sent.

Dates

A date can be represented in MongoDB with ISODate.

{
    "departureTime": ISODate("2019-05-07T09:13:00Z")
    "arrivalDate": ISODate("2019-05-08")
}

We can use filtering based on date, including comparison operators such as $gt and $lt.

Index

To create an index:

db.{collection}.createIndex(
    {
        name: 1
    }
)

The index will be created on the name field in ascending order.

We can get all indexes on a collection with:

db.{collection}.getIndexes()

Mongo supports clustered indexes, so that the primary _id key in the index is directly associated with the BSON. The default _id field in a document is an ObjectId, which takes 12 bytes. It’s quite a lot for a key, but it’s large for uniqueness among shards. Secondary indexes will point to the primary index, which means that secondary indexes become quite big, since each entry of the index needs to contain the 12-byte ObjectId. The situation might be improved (or worsened) with a custom choice of _id.

Tools

Shell

Mongo Shell (mongosh) is a CLI interface for accessing MongoDB. It uses JS-like syntax.

mongosh "mongodb+srv://localhost/MyDatabase" --apiVersion 1 --username "myuser" --password "mypass"

Commands

db // Shows current DB
show dbs // Shows all DBs with storage size
use somedb // Switches context to some DB

show collections // Shows all collections in the current DB
db.createCollection("newcollection") // Creates a new collection
db.{collection}.drop() // Removes a collection

db.{collection}.insertOne({ "name": "Marcin" }) // Creates a document
db.{collection}.insertOne([{ "name": "Marcin" }, { "name": "Tom" }]) // Creates documents

db.{collection}.find() // Gets all documents in a collection

GUI

MongoDB Compass is the official GUI for exploring MongoDB databases. It’s available for Linux, macOS, Windows.

mongoimport

A tool that is added to MongoDB installations that allows for bulk imports of data.

Testing

MongoDB has MongoDB Atlas offering, which is a cloud deploymen of MongoDB. There’s a free tier that is shared, but great for testing. It can be deployed to Azure/GCP/AWS. For testing, we can also load a Sample Dataset.

Sources

• MongoDB Docs
• Pluralsight