Analytics March 2015

White paper

Why NoSQL?
Your database options in the new non-relational world
2 Why NoSQL?

Contents A brief history of NoSQL

NoSQL, also known as not only SQL or non-relational databases,
2 New types of apps are generating new types of data
were specifically introduced to handle the rise in data types,
2 A brief history of NoSQL data access, and data availability needs brought on by the
dot-com boom.
3 Types of NoSQL databases

4 Why consider NoSQL? Going back 20 years or so, when application architects and
developers needed a data store for their applications, they were
5 Summary choosing among various relational databases. In fact, relational
5 Getting started with IBM Cloudant databases have been the de facto choice since the 1970s, as they
have been the sole option available for both developers (taught
5 For more information in computer science programs everywhere) and infrastructure
teams (well-understood tooling and predictable operational
characteristics). Some of the most popular are Oracle, MySQL,
SQL Server and IBM DB2.
New types of apps are generating new
types of data However, when Internet applications and companies started
The continual increase in web, mobile and IoT applications, exploding during the late 90s to early 2000s, applications went
alongside emerging trends shifting online consumer behavior from serving thousands of internal employees within
and new classes of data, is causing developers to reevaluate how companies to having millions of users on the public Internet.
their data is stored and managed. Todays needs require a For these applications, performance and availability were
database that is capable of providing a scalable, flexible solution paramount. The new problem of high availability at large scale
to efficiently and safely manage the massive flow of data to and drove companies like Google, Facebook and Amazon to create
from a global user base. new technologies. Thankfully, they documented their efforts,
released white papers, and open sourced their technology for
Developers and IT alike are finding it difficult, and sometimes the Internet community to continue building upon. By the late
even impossible, to quickly incorporate all of this data into the 2000s, several new non-relational database technologies had
relational model while dynamically scaling to maintain the emerged, and NoSQL was the name that stuck to describe
performance levels users demand. This is causing many to look them all.
at NoSQL databases for the flexibility they offer, and is a big
reason why the global NoSQL market is forecasted to nearly NoSQLs roots in open source
double and reach USD3.4 billion in 20201. Many NoSQL databases have roots in the open source
community. This heritage has been fundamental for their
ever-increasing popularity and usage. You will often see
companies that provide a commercial version of a database
with services and support for the technology, while at the same
time participating in the communities of their open source
 Analytics

counterparts. Examples of these relationships include Datastax

for Apache Cassandra, IBM Cloudant for Apache CouchDB,
and even MongoDB with an open source version of its
MongoDB software.

Types of NoSQL databases

NoSQL is simply the term used to describe a family of
databases that are all non-relational. While the technologies,
data types, and use cases vary wildly among them, it is
generally agreed that there are four types of NoSQL databases:

Key-value stores
These databases pair keys to values, like a hash table in
computer programming. A good analogy for a key-value store
is a file system. The path acts as the key and the contents act as
the file. There are often no fields to update; rather, the entire run range queries over a specific value. Examples of column stores
value other than the key must be updated if changes are to be include Hbase and Cassandra.
made. This simplicity scales well; however, it can limit the
complexity of queries and other advanced features available in Document stores
key-value stores. Examples of pure key-value stores include Document databases store records as documents, where a
MemcacheD, REDIS and Riak. document can generally be thought of as a grouping of
key-value pairs. Keys are always strings and values can be
Graph stores stored as strings, numerics, booleans, arrays and other nested,
Graph data stores excel at dealing with interconnected data. key-value pairs. Values can be nested to arbitrary depths.
Graph databases consist of connections (called edges in graph Popular syntaxes for document DBs include XML and
DB terms) between nodes. Both nodes and their edges can JavaScript Object Notation (JSON). Examples of document stores
store additional properties, such as key-value pairs. The include MongoDB, CouchDB, Cloudant and MarkLogic.
strength of a graph database is in traversing the connections
between nodes. Graph databases, however, generally require all Why consider NoSQL?
data to fit on one machine, limiting their scalability. Examples The various NoSQL databases available today differ quite a
of graph stores include Neo4j and Sesame. bit, but there are common threads uniting them: flexibility,
scalability, availability, lower costs and special capabilities.
Column stores
Relational databases store all of the data in a particular tables Flexibility
row together on disk, which makes retrieval of a particular row Schema flexibility and intuitive data structures are key features
fast. Column-family databases generally serialize all the values
of a particular column together on disk, which makes retrieval
of aggregated data on a specific attribute fast. This is valuable
in data warehousing and analytics scenarios where you might
4 Why NoSQL?

that attract developers working in agile development cycles,

and most NoSQL databases accommodate these qualities. The Appeal of NoSQL JSON document stores
Schema updates and their associated downtime are generally JSON schema can be evolved rapidly without the need for intervention
not required, unlike in a relational database. NoSQLs It only contains six kinds of values, so it is easy to implement and use
JSON is self-describing and easy to understand
flexibility is popular for supporting agile development practices
It helps enable performance and scalability gains
by eliminating the complexity of database schema changes to
support changing codebases. Example: NoSQL JSON document store versus a RDBMS
Records in a NoSQL document store may be fully denormalized into
Scalability individual JSON documents (store a record and all its related data together)
Scale refers to both the size of the data as well as the versus the RDBMS approach of decomposing relations into separate
tables (separate records into logical tables to better enforce consistency).
concurrent users acting on that data. NoSQL databases are
The benefit here is that by using NoSQL, you can get and put objects at
typically more specialized to various use cases involving scale
once without JOINing data from separate tables. The JOINs of a relational
and can be much simpler for developing related application database are absolute scalability killers for clustered databases.
functions than relational databases.

Many NoSQL databases are built to scale horizontally and

of NoSQL databases means that they can be deployed and
spread their data across a cluster of servers more easily than
operated on clusters of servers in cloud architectures.
their relational counterparts. Relational database performance
suffers when queries execute JOINs across shards, whereas a
Specialized capabilities
NoSQL database can avoid JOINs altogether and maintain
Not all NoSQL databases are created equal. To offer incentive
high performance.
and added integration to their users, many NoSQL providers
include various specialized capabilities. Examples include
Availability
specific indexing and querying capabilities for geospatial data,
Downtime results in lost revenue, lost customers, and
integrated full-text indexing for search, automatic data
frustrated users. Thankfully, some NoSQL databases offer new
replication and synchronization features and application-
or different replication architectures that can make different
friendly RESTful web APIs.
types of NoSQL databases more highly available for writes in
addition to reads. This means that if one or more database
servers, or nodes, goes down, the other nodes in the system are Summary
able to continue operations without data loss. The database you pick for your next application matters now
more than ever. Thankfully, you dont have to pick just one.
Lower operational cost Todays applications are expected to run non-stop and must
The open source roots of NoSQL make its entry point an efficiently manage continuously growing amounts of multi-
obvious incentive, and it is common to hear of NoSQL structured data in order to do so. This has caused NoSQL to
adopters cutting significant costs versus their existing databases grow from a buzzword to a serious consideration for every
while still receiving the same or better performance and
functionality. Historically, large relational databases have run
on expensive machines and mainframes. The distributed nature
 Analytics

database, from small shops to the enterprise.

While NoSQL databases share some common qualities, such

as being non-relational and generally easy to scale, there are
many unique differentiators to consider when deciding upon
the correct NoSQL solution for your unique data needs. The
right NoSQL database can act as a viable alternative to
relational databases or can be utilized in a complementary
fashion along with existing systems.

The requirement for efficient data delivery is our future, and

you should consider NoSQL technology when planning any
application development project that involves scale, different
varieties of data or large amounts of potential users.

Getting started with IBM Cloudant

IBM Cloudant is available as a fully managed NoSQL database
as a service (DBaaS) for fast, turnkey provisioning and worry-
free data management. It is also available as Cloudant Local,
which puts the power of the Cloudant platform in the privacy of
your data centers. You can even connect Cloudant Local and
Cloudant Managed DBaaS databases together to form hybrid
cloud databases for the greatest balance of cloud cost, reach,
performance, and compliance control. Simply sign up for a
no-charge account and get started at cloudant.com.

For more information

For more information, please contact your IBM representative
or IBM Business Partner, or visit cloudant.com or
ibm.com/cloudant.
 Copyright IBM Corporation 2015

IBM Corporation
Software Group (or appropriate division, or no division)
Route 100
Somers, NY 10589

Produced in the United States of America

February 2015

IBM, the IBM logo, and ibm.com are trademarks of International Business
Machines Corp., registered in many jurisdictions worldwide. The Apache
Software Foundation, Apache Cassandra, CouchDB, and Apache
CouchDB are trademarks or registered trademarks of The Apache
Software Foundation. IOS is a trademark or registered trademark of Cisco
Systems, Inc., in the U.S. and other countries. Other product and service
names might be trademarks of IBM or other companies. A current list of
IBM trademarks is available on the Web at Copyright and trademark
information at www.ibm.com/legal/copytrade.shtml.

This document is current as of the initial date of publication and may be

changed by IBM at any time. Not all offerings are available in every country
in which IBM operates.

THE INFORMATION IN THIS DOCUMENT IS PROVIDED

AS IS WITHOUT ANY WARRANTY, EXPRESS OR IMPLIED,
INCLUDING WITHOUT ANY WARRANTIES OF MERCHANT
ABILITY, FITNESS FOR A PARTICULAR PURPOSE AND ANY
WARRANTY OR CONDITION OF NONINFRINGEMENT. IBM
products are warranted according to the terms and conditions of the
agreements under which they are provided.

1 www.marketresearchmedia.com/?p=568

Please Recycle

KUW12354-USEN-00