column stores

• data oriented by columns, not rows

key-value stores

• schema-less data stored by keys

document stores

• document-oriented data stored by keys

graph databases

• graph structures with nodes, edges, and properties

tuple stores, object databases, many others

NoSQL databases deal with Big Data challenges:
velocity, volume, variety

scale horizontally with demand, fits with cloud oriented architecture

support the needs of modern web applications

• BSON/JSON encoding
• HTTP REST interfaces

distributed architectures

• what happens when your app loses its internet cxn?

event driven architectures

• how does your application know when its data is stale?

store documents addressed by keys

documents might be encoded in JSON, BSON, XML, YAML, or could possibly be binary formats like PDF, Microsoft Word, Microsoft Excel, etc.

JSON encoding with a RESTful HTTP API is popular

• JSON and REST are easy to work with in the context of a browser
• JSON maps directly to primitive data types and collections
• REST is HTTP, so caching, load-balancing, and other techniques easily apply to the database

REpresentational State Transfer

• Roy Fielding, Architectural Styles and the Design of Network-based Software Architectures

describes the design rationale behind the architecture of the web as a set of constraints

• client-server, stateless, cache, uniform interface, layered system, code-on-demand

in the context of HTTP, it describes how well-designed web applications behave

a web API implemented using HTTP and REST principles

collection URI: http://www.ucalgary.ca/it/

• GET: list the URI
• PUT: replace the collection with another collection
• POST: create a new entry in the collection
• DELETE: delete the entire collection

element URI: http://www.ucalgary.ca/it/banner.jpg

• GET: retrieve a representation of the member
• PUT: replace the member
• POST: generally not used
• DELETE: delete the member

Javascript Object Notation

derive from JavaScript, but is langauge-independent

data types

• numbers 1 2.5
• strings åbçdé
• booleans true false
• arrays [1, 2, 3]
• dictionaries {“key”: “value”}
• null

document-oriented databases

string keys and JSON documents

HTTP REST based APIs

event-based change notifications

• use for live updates, data sychronization

scalable across multiple nodes

support client-side web applications

started in 2005, accepted into Apache Incubator in 2008

uses MapReduce to generate views

has additional functionality to support client-side web applications

• documents can have attachments
• _design documents can have view, show, list, update, and filter functions to control resource repr. and partitioning
• _changes feed can provide real-time notifications of changes in a database

we use CouchDB to synchronize ADSM with Remedy

Apache CouchDB

do something with Firebase and/or CouchDB

Firebase suggestions

• take one of the examples and expand on it

CouchDB suggestions

• create MapReduce views against CM or Remedy data
• write a show function for CM or Remedy data
• write a client-side to-do list app (use jquery.couch.js)
• adapt a Firebase example

we’ll stop at 10:50 to share results

at 11:15, Chris Ismael (Mashape) will be talking about Cloud APIs and machine learning