Title Text

SQL.

It's so last century, amirite?

'NoSQL'

Web-trending since ~2009

'SQL alternative'

Book-trending since ~1984

Why NoSQL?

The reasons are legion...

Impedance Mismatch

UI

DB

Object Hierarchies

Table Relationships

Differing data representations

on client and server

Clustering

Does this really scale?

Alternative Approaches

Key/Value Pair

Column Family

Document

Graph

Cassandra

Apache HBASE

Oracle Berkley DB

Redis

CouchDB

MongoDB

Firebase

Riak KV

Neo4J

Sqrrl

GraphDB

ScyllaDB

Key/Value Pair

Oracle Berkley DB

Redis

Riak KV

  • Value could be any type (string, number, blob, etc.)
  • Some support ordering of keys
  • Many are in-memory implementations for speed
  • Takes up less memory than most other types since relationships are simplest¬†
  • More implementations than any other NoSQL approach

Document

CouchDB

MongoDB

Firebase

  • Values are either lists of key/value pairs or 'documents'
  • Similar to KVP databases
  • Documents can be modified whole or in part
  • Some implementations support secondary indexing
  • Most commonly JSON, eliminating impedance mismatch

Column Family

Cassandra

Apache HBASE

ScyllaDB

  • Similar to SQL in use of 'rows'
  • Similar to Document DB in that sections can be addressed
  • As with most other NoSQL approaches, notion of schema is flexible / dynamic

Graph

Neo4J

Sqrrl

GraphDB

  • Radically different from other NoSQL approaches
  • Data model is based on nodes and edges rather than aggregation
  • While both are 'relational', graph databases are more optimized than SQL for navigating and retrieving data by relationship

ACID

BASE

Atomicity (of updates), Consistency (of state), Isolation (of users from in-progress updates), Durability (of transactions across system failures)

Characteristics of SQL vs NoSQL Systems

Basically Available (data is never 'locked'), Soft state (system can change over time even without input, because of...), Eventually consistent (once input stops)

CAP Theorem

Availability

Consistency

Partition Tolerance

Clients can always read & write

Clients always see same view of data

Clients never affected by partitioning faults / delays

Choose

any

2

SQL / RDBMS
Disk-based KVP DBs
Some Document DBs
In-memory KVP DBs
Some Column DBs
Some Column DBs
Some Document DBs
Some Column DBs

Resources

Introduction to NoSQL (Martin Fowler)

https://youtu.be/qI_g07C_Q5I

What is a Document Database?

https://aws.amazon.com/nosql/document/

Don't Believe The Hype, It's a SQL

By Clifford Hall

Don't Believe The Hype, It's a SQL

SQL. It's so 1974, amirite?

  • 488
Loading comments...

More from Clifford Hall