1.
SQL vs. NoSQL
Making the right choice
18 September 2014

2.
The contenders
SQL NoSQL
© Copyright Dimension Data 18 September 2014 2

3.
SQL Databases
• RDBMS
• Standardized
• Mature
• Reliable
• Well understood
• Queryable
• ACID
© Copyright Dimension Data 18 September 2014 3

4.
NoSQL scalability argument
• Scale-Up vs Scale-Out
• Use of commodity hardware
• Locking / Latching
• Consistency over partitions
• Availability of partitions
• Referential integrity
Cost of scaling
SQL NoSQL
© Copyright Dimension Data 18 September 2014 4

5.
Other RDBMS / SQL Database drawbacks
• One-solution-fits-all
• Slow for certain tasks
• ACID is not always needed
• ORM required
• Lack of flexibility
• Rigid schema
• Management complexity
• Add-on solutions
• XML-fields, Filestreams
• Full-text indexes
© Copyright Dimension Data 18 September 2014 5

6.
CAP theorem (Brewer's theorem)
© Copyright Dimension Data 18 September 2014 6

7.
NoSQL Use Cases
• Bigness / Avoid hitting the wall
• Massive write performance
• Write availability
• Fast key-value access
• Flexible schema and flexible datatypes
• Schema migration
• No single point of failure
• Generally available parallel computing
• Easier maintainability, administration and operations
• Programmer ease of use
• Use the right data model for the right problem
• Tunable CAP tradeoffs
© Copyright Dimension Data 18 September 2014 7

8.
ACID Transactions
Atomicity
Consistancy
Isolation
Durability
© Copyright Dimension Data 18 September 2014 8

9.
NoSQL ACID Trade-offs
• Dropping Atomicity lets you shorten the
time tables (sets of data) are locked.
MongoDB, CouchDB.
• Dropping Consistency lets you scale up
writes across cluster nodes.
Riak, Cassandra.
• Dropping Durability lets you respond to
write commands without flushing to disk.
Memcache, Redis.
© Copyright Dimension Data 18 September 2014 9

10.
NoSQL Database Main Types
• Key-Value Store
• A basic dictionary design storing values under unique keys
• The database does not care about the structure of the value
• Examples:
• Memcache
• Riak
• Azure Blob Storage
• Good at:
• Handles size well
• Processing a constant stream of small reads and writes
• Fast
• Programmer friendly
© Copyright Dimension Data 18 September 2014 10

11.
NoSQL Database Main Types
• Column Store
• A column is a tuple of 3 elements: unique name of value, a typed
value, timestamp
• Columns may be part of column families
• Columns need not appear in every record
• Example:
• Hbase
• Hypertable
• Cassandra
• Azure Table Storage
• Good at:
• Handles size well
• Stream massive write loads
• High availability
• Multiple-data centers
• MapReduce.
© Copyright Dimension Data 18 September 2014 11

12.
NoSQL Database Main Types
• Document Store
• Use a unique key to store and retrieve a JSON document
• Documents are schemaless
• Metadata is added to the document to aid querying
• Indexing of documents and metadata speeds up retrieval
• Example:
• CouchDB
• MongoDB
• RavenDB
• Azure DocumentDB service (Preview)
• Good at:
• Natural data modeling
• Programmer friendly
• Rapid development
• Web friendly
• CRUD
© Copyright Dimension Data 18 September 2014 12

13.
NoSQL Database Main Types
• Graph Database
• Uses graph structures with nodes, edges, and properties to represent
and store data
• Every element contains a direct pointer to its adjacent elements
• Example:
• AllegroGraph
• InfoGrid
• Neo4j
• Good at:
• Complicated graph problems
• Topographical data
• Fast
© Copyright Dimension Data 18 September 2014 13

14.
NoSql Database Type Comparison
Data Model Performance Scalability Flexibility Complexity Functionality
Key–Value
Store
high high high none variable (none)
Column-
Oriented Store
high high moderate low minimal
Document-
Oriented Store
high variable (high) high low variable (low)
Graph
Database
variable variable high high graph theory
Relational
Database
variable variable low moderate
relational
algebra
© Copyright Dimension Data 18 September 2014 14

15.
Things to consider when choosing
• Where are you starting from?
• What are you trying to accomplish?
• Things to Consider...
• Your Problem
• Access pattern, scalability, consistency, durability
• Money
• Scaling, admins, license, operating cost
• Programming
• Flexible schema, JSON, REST, language, graphs
• Performance
• Reads, writes, consistency, workload, eventual consistency
• Features
• Cross datacenter, upgrades, indexes, persistence, tunability
• The vendor
• Viability, future direction, responsiveness, partnerships
© Copyright Dimension Data 18 September 2014 15

16.
Big Data – Petabyte range
Microsoft HDInsight
=
Hadoop as a service on Azure (+ .NET)
© Copyright Dimension Data 18 September 2014 16

17.
Hadoop components
© Copyright Dimension Data 18 September 2014 17

18.
Using Hadoop
© Copyright Dimension Data 18 September 2014 18

19.
Hadoop cluster size
Yahoo! wins with a massive 42000 node cluster
© Copyright Dimension Data 18 September 2014 19

20.
Questions
USE [Euricom]
SELECT [Question]
FROM [dbo].[FAQ]
WHERE [Answer] IS NULL
(0 row(s) affected)
© Copyright Dimension Data 18 September 2014 20