MongoDB

Reference

Summarise from https://www.tutorialspoint.com/mongodb

Overview

• Server / client - mongod / mongo
• Database ("test" the default)
  • physical container for collections
  • get own set of files on file system
  • single server -> many dbs
• Collection (like a RDBMS table)
  • group of documents
  • exists within a single db
  • do not enforce a schema
  • can have different fields
  • usually all docs in one collection have similar or related purposes
• Document (like a RDBMS row)
  • set of key/value pairs (columns)
  • have dynamic schema
    • docs in same collection can have different set of fields and structure
    • common fields in docs can hold different type of data
  • embedded documents (RDBMS table join)
  • Primary key - default key _id provided by mongoDb
    • _id:ObjectId(xxxxxxxxxxxx) - 12 bytes hexadecimal number, unique for every doc (4 bytes timestamp, 3 machine id, 2 process id, 3 serial)
  • JSON / BSON

Vs. RDBMS

https://halls-of-valhalla.org/beta/articles/the-pros-and-cons-of-mongodb,45/

Pros:

• schema-less
• no complex joins
• deep query-ability, document-based query language
• no ORM
• easy sharding (partitioning)

Cons:

• No joins
• No schema - store key names in memory - memory usage high - slow query
• Concurrency issues
• Young
• Transactions - manual

Environment

Start/stop: sudo service mongod start/stop

Shell: mongo

Help: db.help()

Data Modelling

In same collection data has a flexible schema.

Considerations:

• Combine objects into one document if they are used together
• Otherwise separate objects but make sure they should not need joins
• Duplicate data (be careful!) as disk space is cheaper than compute time
• Do joins while write / not on read
• Optimise for most frequent use cases
• Do complex aggregation in schema

DB Operations

create or return to: use DB_NAME

check current db: db

list db (need at least one document): show dbs

use db.[collection].insert() to insert document

drop selected db: db.dropDatabase()

Collection Operations

db.createCollectin(name, options) - options are optional

• capped - fixed sized collection, automatically overwrite oldest entries
• autoIndexed - automatically create _id, default false

>db.createCollection("mycol", { capped : true, autoIndexId : true, size : 

   6142800, max : 10000 } )

drop: db.COLLECTION_NAME.drop()

list: show collections

Data types

• String - UTF8
• Integer - 32/64bit depending upon server
• Boolean
• Double
• Min/Max keys - compare a value against the lowest/highest BSON elements
• Arrays
• Timestamp
• Object - for embedded documents
• Null
• Symbol - used identically to a String, however reserved for languages that use a specific symbol type
• Date - used to store UNIX time format, can specify own data time by creating object of data
• Object ID
• Binary data
• Code - javascript code
• Regular expression

Document operation

db.COLLECTION_NAME.insert(document)

>db.mycol.insert({

   _id: ObjectId(7df78ad8902c),

   title: 'MongoDB Overview', 

   description: 'MongoDB is no sql database',

   by: 'tutorials point',

   url: 'http://www.tutorialspoint.com',

   tags: ['mongodb', 'database', 'NoSQL'],

   likes: 100

})

(if COLLECTION not exist, will be created)

Can pass in an array of documents in insert() command

db.post.save(document)

• if no _id then save() will work same as insert()
• if specify _id it replace whole data of document

Query

db.COLLECTION_NAME.find()

- prints all documents in non-structured way

db.COLLECTION_NAME.find().pretty()

- prints in formatted way

db.COLLECTION_NAME.findOne() - return just one

The above accepts JSON objects as conditon, see: https://www.tutorialspoint.com/mongodb/mongodb_query_document.htm

not a pretty language as SQL, best use some visual tools

Update/Save/Remove

Update method accepts SELECTION criteria (JSON) and UPDATED DATA criteria (JSON), will update a single record only unless specified {multi:true}

>db.mycol.update({'title':'MongoDB Overview'},

   {$set:{'title':'New MongoDB Tutorial'}},{multi:true})

save() method replaces existing document with new one

>db.COLLECTION_NAME.save({_id:ObjectId(),NEW_DATA})

remove()

>db.mycol.remove({'title':'MongoDB Overview'}) // remove all selected

>db.COLLECTION_NAME.remove(DELETION_CRITERIA,1) // remove just one, or the first

>db.mycol.remove() // remove all

Projection (select partial results, sort of results)

>db.mycol.find({},{"title":1,_id:0}) // select title but hide _id

>db.COLLECTION_NAME.find().limit(NUMBER) // limit number of documents

>db.COLLECTION_NAME.find().limit(NUMBER).skip(NUMBER) // skip a number of documents

>db.COLLECTION_NAME.find().sort({KEY:1}) // sort, 1 ascending (default) -1 descending

Index

>db.mycol.ensureIndex({"title":1,"description":-1}) // index key by ascending / descending order

More options:

• background - build in background so not blocking activity
• unique
• name - name of index
• dropDups - drop second and more documents when creating index
• sparse ??
• expireAfterSeconds - TTL, control how long DB retains documents in this collection
• v - index version, default depends on DB version
• weights - significant of field relative to other indexed fields (??)
• default_language - string, relate to stop words and rules for stemmer and tokenizer. for text index
• language_override - name of the field in the document that contains the language (to override default language)

Aggregate

>db.COLLECTION_NAME.aggregate(AGGREGATE_OPERATION)

> db.mycol.aggregate([{$group : {_id : "$by_user", num_tutorial : {$sum : 1}}}])

• $sum : value - sum up value (value can be $field_name, for example)
• $avg
• $min / $max
• $push - insert value to an array in the resulting document
• $addToSet - insert to a set
• $first / $last

Pipeline Concept