by Oscar Ivarsson

MongoDB

Show-Off

What is it?

Type: Document Database

Designed as a scalable database (humongous)

Open Source

Released: 2009

Notation

Database

Collection

RDBMS

Schema

Instance

MongoDB

Instance

Table

Document

Row

Example

> printjson(db.towns.findOne({"_id":ObjectId("4d0b6da3bb30773266f39fea")}))
{
    "_id":ObjectId("4d0b6da3bb30773266f39fea"),
    "country": {
        "$ref": "countries",
        "$id": ObjectId("4d0e6074deb899526a8309e")
    },
    "famous_for": [
        "beer",
        "food"
    ],
    "last_census": "Thu Sep 20 2007 00:00:00 GMT    -0700 (PDT)",
    "mayor": {
        "name": "Sam Adams",
        "party": "D"    
    },
    "name": "Portland",
    "population": 582000,
    "state": "OR"
}

Stored as BSON  - Shown as JSON

Database: "db"
Collection: "towns"
Document: Between the brackets
Identifier: "_id"

0-3: Timestamp
4-6: Client Machine ID
7-8: Client Process ID
9-11: Incremented Counter

Notations

ObjectId (12 bytes)

Examples

> db.test.insert({color: "Red"})
> db.test.insert({shape: "Triangle"})

Insert

> db.test.find()
{ 
    "_id" : ObjectId("55ee924b34ec1a858b481962"),
    "color" : "Red"
}
{
    "_id" : ObjectId("55ee927f34ec1a858b481963"),
    "shape" : "Triangle"
}

Find

Examples

> db.test.find({ "_id" : ObjectId("55ee927f34ec1a858b481963") })
{ 
    "_id" : ObjectId("55ee927f34ec1a858b481963"),
    "shape" : "Triangle"
}

Find by ID

> db.test.find( { "shape": "Triangle"} )
{ 
    "_id" : ObjectId("55ee927f34ec1a858b481963"),
    "shape" : "Triangle"
}

Find by field (slower..)

> db.test.find( function() {
    return this.shape == "Triangle";
  })

{ 
    "_id" : ObjectId("55ee927f34ec1a858b481963"),
    "shape" : "Triangle"
}

Find by function (slowest!)

Query Commands

Examples for find

Examples for update

References

Mongo is not built for performing joins because of its distributed nature!

Documents can however reference eachother.

> db.towns.update(
    { _id : ObjectId("4d0ada87bb30773266f39fe5") },
    { $set : { country: { $ref: "countries", $id: "us" } } }
  )

THEN

> var portland = db.towns.findOne({ _id : ObjectId("4d0ada87bb30773266f39fe5") })
> db.countries.findOne({ _id: portland.country.$id })

OR

> 
db[ portland.country.$ref ].findOne({ _id: portland.country.$id })

Performance

Query performance can be increased in MongoDB by using indexing.

Full collection scan

B-tree indexing

Scanned objects: 1 000 000

Scanned objects: 1

Execution Time 1st: 470 ms

Execution Time 2nd: 365 ms

Execution Time 1st: 5 ms

Execution Time 2nd: 0 ms

> populatePhones(800, 809, 5550000, 5650000)
> db.phones.find().limit(1)

{ 
    "_id" : 18005550000, 
    "components" : { 
        "country" : 1,
        "area" : 800,
        "prefix" : 555,
        "number" : 5550000 },
        "display" : "+1 800-5550000" 
}

Scaling Out!

What makes document databases unique?

Their ability to efficiently handle arbitrary nested, schemaless data documents.

What makes MongoDB special?

The ability to scale across several servers by replicating or sharding collections.

MongoDB was built to scale out.

When to use it?

Strengths

Weaknesses

Replication and horizontal scaling

Flexible data model

The price for flexibility -> Error feedback

Development costs and speed

No joins of collections

Availability

Use Cases

Real-Time Analytics

City of Chicago

McAfee

Crittercism

Use Cases

Product Data Management

Otto

Under Armour

Carfax

Use Cases

CMS

Forbes

Ebay

Pearson

http://db-engines.com/en/ranking