•

what is MongoDB

•

non-relational

•

doesn't store in tables

•

stores in JSON documents

•

key/value

•

can also be key/array

•

and key/array(key/...)

•

MongoDB is a JSON document store

•

MongoDB data looks a lot more like the data you have in your programs than relational data does

•

schemaless

•

two documents don't need to have the same schema as two SQL rows do

•

where does MongoDB fit in the world?

•

two axis:

•

scalability and performance

•

memcached

•

scalable and has good performance, but offers little functionality

•

depth of functionality

•

Oracle

•

DB2

•

SQL Server

•

not known for being particularly scalable: easy to scale up but not easy to scale out (use commodity hardware)

•

MongoDB strikes the balance between these two

•

what is missing from MongoDB which classic RDBMS have?

•

no joins

•

each document is stored in a collection

•

the reason is: joins is one of the items which scales particularly poorly when you try to scale out

•

no transactions

•

sounds bad, but you often don't need them in applications which use MongoDB

•

because documents are hierarchicals, you get some of this functionality

•

and it also is the case in real-world RDBMS scenarios that you do not have transactions between separate databases anyway

•

what is included in MongoDB which RDBMS have?

•

indexes

•

secondary indexes

•

although these are challenging to get them to perform well on a system which is horizontally scalable

•

browser --> layout engine --> JavaScript engine

•

Firefox --> Gecko --> SpiderMonkey

•

Safari --> WebKit --> Nitro

•

Chrome --> WebKit --> V8

•

2013: Chrome --> Blink --> V8

•

Opera --> Presto --> Carakan

•

2013: Opera --> Blink --> V8

•

Internet Explorer --> Trident --> Chakra

•

layout engines

•

Gecko

•

Firefox

•

KHTML

•

Konqueror

•

Trident

•

Internet Explorer

•

WebKit

•

Safari

•

Google Chrome (2013: discontinued using it)

•

Presto

•

Opera (2013: phased it out)

•

Blink

•

Chrome

•

April 2013: Google forked from WebKit

•

even before the split, Chrome never used WebKit in same way as Safari, e.g. Chrome ignored WebKit's JavaScriptCore and used V8 instead

•

Opera

•

July 2013: Opera switched to Blink

•

Opera 15+

•

Webview (Android)

•

as of 4.4

•

JavaScript engines

•

SpiderMonkey

•

1995: Brendan Eich, Netscape

•

C/C++

•

SpiderMonkey name has remained the same but modern engine is unrecognizable from 1995 engine

•

JIT compilers for SpiderMonkey

•

TraceMonkey

•

JägerMonkey

•

IonMonkey (current)

•

translates SpiderMonkey bytecode into a control flow graph

•

default engine in Firefox since version 18

•

Rhino

•

1997: Netscape

•

developed entirely in Java

•

separate from the SpiderMonkey

•

Apple Safari 4's Nitro

•

Google Chrome's V8 engine

•

converts JavaScript into classes

•

works in compiled and interpreted mode

•

intended for server-side applications

•

no built-in support for web browser objects

•

JavaScriptCore

•

Apple forked KJS (Konqueror) to create JavaScriptCore for WebKit (layout engine)

•

2008 WebKit gets SquirrelFish

•

SquirrelFish Extreme, Nitro, Nitro Extreme, compiles JavaScript directly to machine code (like V8)

•

JavaScriptCore source code resides in the WebKit source tree

•

V8

•

2008 with Chrome 2, in response to Nitro

•

compiles JavaScript to native machine code

•

intended to be used both in a browsers, e.g. Chrome and Chromium (open source web browser project from which Google Chrome draws its source code, the browsers share the majority of code and features)

•

goal for Chrome is to be a "a tabbed window manager, or shell for the web, as opposed to it being a traditional browser application"

•

Chakra

•

as of Internet Explorer 9

•

uses GPU for 3D graphics and video

•

MongoDB from 10,000 feet

•

clients

•

web browsers

•

any client sending requests, e.g. consuming REST API

•

application server

•

running node.js

•

C++ program that you control using V8 JavaScript

•

all code for it are written in JavaScript

•

server responds to requests from clients

•

this application needs to store persistent data, that is where MongoDB comes in

•

MongoDB server

•

written in C++

•

application server acts as client and communicates with MongoDB server which is listening for requests

•

Mongo Shell

•

has similarities with node.js

•

C++ application that you control with V8

•

makes requests to MongoDB

•

looking at the data

•

debugging your application

•

the driver

•

handles all the connections, fail over

•

provides API which handles the communication to MongoDB