MongoDB 101 

I am... • Nick Jackson • Awesome Developer Dude • University of Lincoln • @jacksonj04 

MongoDB Is... • A NoSQL document database. 

Eh? 

NoSQL • Is not a specification (unlike SQL). • (Generally) doesn’t have schemas. • (Generally) works at web-scale. • (Generally) doesn’t do relational integrity. 

Document Databases • Store documents. • Don’t store key-value pairs. • Make building APIs and stuff really easy. 

Ye Olde SQL id name email 1 Nick [email protected] 2 Joss [email protected] 

Ye Olde SQL id name email office building 1 Nick ... 3107 MHT 2 Joss ... 3015 MHT 3 Harry ... NULL MHT 

:-( 

Documents: Awesome! { name: ‘Nick’, email: ‘[email protected]’, location: { office: ‘3105’, building: ‘MHT’ } } 

Documents: Awesome! { name: ‘Harry’, email: ‘[email protected]’ } 

Documents: Awesome! { name: ‘Joss’, email: ‘[email protected]’, location: { office: ‘3105’, building: ‘MHT’ }, likes: { music: [‘Folk’, ‘Hip-Hop’], drink: [‘Coffee’, ‘Ale’] } } 

Servers Are Easy 

On Its Own Mongo App 

Replicated Mongo 1 Mongo 2 Mongo 3 App 

Sharded App Router 1 Router 2 Mongo S1 Mongo S2 Mongo S3 

http://mongodb.org 

Inserts <3 JSON > db.people.save({name:'Nick'}) > db.people.save({name:'Joss'}) > db.people.save({name:'Harry'}) 

Query Be Simple... > db.people.find() { "_id" : ObjectId("4fa...103"), "name" : "Nick" } { "_id" : ObjectId("4fa...104"), "name" : "Joss" } { "_id" : ObjectId("4fa...105"), "name" : "Harry" } 

Query Be Simple... > db.people.find({name:'Nick'}) { "_id" : ObjectId("4fa...103"), "name" : "Nick" } 

Query Be Quick... > db.people.ensureIndex({name:1}) 

Updates are easy > db.people.update({name:'Nick'},{name:'Nick',likes: ['coffee']}) > db.people.find({name:'Nick'}) { "_id" : ObjectId("4fa...103"), "name" : "Nick", "likes" : [ "coffee" ] } 

Queries are powerful > db.people.update({name:'Joss'},{name:'Joss',likes: ['coffee','folk music']}) > db.people.find({likes:'coffee'}) { "_id" : ObjectId("4fa...103"), "name" : "Nick", "likes" : [ "coffee" ] } { "_id" : ObjectId("4fa...104"), "name" : "Joss", "likes" : [ "coffee", "folk music" ] } 

Deletes are also easy > db.people.remove({name:'Nick'}) > db.people.find() { "_id" : ObjectId("4fa...104"), "name" : "Joss", "likes" : [ "coffee", "folk music" ] } { "_id" : ObjectId("4fa...105"), "name" : "Harry" } 

Cool Things! • Geospatial indexes. db.places.find({loc:{$near:[-2, 53]}}) • JavaScript in the Mongo shell. • Map/Reduce operations. • Can be used as a filesystem. 

Downsides It has a few 

It’s not ACID • Set of atomic operators, but no things like transactions. • No enforced consistency. At all. • No locking, so updates can collide and be lost. • Disk writes are (usually) deferred, so data can be lost in failures. 

It’s not ‘Enterprise’ • Your DBAs will find it new and scary. • You need to un-learn a lot of the SQL mindset. • It’s not seen as ‘proven’, but this is generally rubbish. 

Some MongoDB Users • Craigslist • MTV • SourceForge • Disney • National Archives • HM Government • The Guardian • New York Times • bit.ly • GitHub • Foursquare • http://lncn.eu/fhx5