WHITEPAPER

Why NoSQL
Databases?
Why successful projects rely on
NoSQL database applications
What is NoSQL?

NoSQL is a modern database management system Relational vs. NoSQL: What’s the
that stores information in JSON documents instead of
difference?
the columns and rows used by relational databases.
Relational databases were born in the era of mainframes
It can deliver data for applications in ways that make
and back-office business applications – long before the
development easier and more robust.
internet, the cloud, big data, mobile, etc. In fact, the first
NoSQL databases are built from the ground up to be
commercial implementation was released by Oracle in
flexible, scalable, and capable of rapidly responding to the
1979. These databases were engineered to run on a single
data management demands of modern businesses.
server – the bigger, the better. The only way to increase
This paper introduces the modern challenges that NoSQL the capacity of these databases was to upgrade the
databases address and shows when to choose NoSQL over server – processors, memory, and storage – to scale up as
relational and why. Moore’s Law allowed.

NoSQL databases emerged as a result of the exponential

Customer experience drives enterprise growth of the internet and the rise of web applications.
to NoSQL solutions Google Bigtable research was released in 2006, as well
as an Amazon Dynamo research paper in 2007. Efficient
Customer experience is the most important competitive
distributed key-value (KV) engines were essential to this
differentiator and businesses are working to meet these
evolutionary step and have propelled the technology
new expectations with services that are on demand, real
much further.
time, resilient, and responsive.
New databases were engineered to meet the next
Bringing together all of these new systems requires
generation of enterprise requirements, which companies
flexibility, performance, and scalability and can consolidate
like Couchbase have taken even further to meet needs
multiple types of systems to reduce database sprawl.
going into the future – the need to develop with agility
and to operate at any scale.

Agility means providing flexible schemas, APIs, robust

SQL-based querying, text search, analytics, and more.
h
e wit Scalability allows data to grow without sacrificing
replac
performance and stability.
h
e wit
replac

2 vCPU 4 vCPU 8 vCPU

15Gib RAM 30Gib RAM 61Gib RAM
475mb 950mb 1900mb

WHITEPAPER 2
Supporting SQL
and NoSQL developers
Traditional relational systems manage tabular data and
return it as rows and columns. NoSQL databases can do
this without forcing application developers into using a
static schema that has to be reworked every time there is
a change. Instead, NoSQL databases give developers the
flexibility they need to help them excel at their work.

NoSQL systems hold hierarchical JSON data but return it

to the application as full or partial JSON data structures,
full-text search matches, SQL query results, key-based
values, or even big data analytics systems.

This convergence of the best of relational and the best of

modern NoSQL simplifies the information architecture of
enterprises and helps developers deliver applications more
efficiently with familiar concepts and tooling, without
needing to learn a dozen different platforms.

Scaling beyond SQL databases

To operate at scale, NoSQL systems approach cluster- SQL-based relational databases are unable to meet these
based computing with efficient, automatic cluster requirements efficiently and cost-effectively.
management to keep data synchronized and flowing at
Consider just a few examples of Global 2000 enterprises
high speeds.
that are deploying NoSQL for mission-critical applications
For example, teams are now expected to build enterprise that have been featured in recent news reports:
data management infrastructure that includes the • Tesco, Europe’s No. 1 retailer deploys NoSQL for
following characteristics: e-commerce, product catalog, and other applications

• Support large numbers of concurrent users (tens of • Ryanair, the world’s busiest airline uses NoSQL to
thousands, perhaps millions) power its mobile app serving over 3 million users

• Deliver highly responsive experiences to a globally • Marriott hosts 30 million documents, accessed at
distributed base of users 4,000 transactions per second

• Be always available – no downtime • GE deploys NoSQL for its Predix platform to help
manage the industrial internet
• Handle semi- and unstructured data
• Sky / Peacock deploys NoSQL to offer a seamless
• Rapidly adapt to changing requirements with
viewing experience during peak watching times
frequent updates and new features

WHITEPAPER 3
Today’s trends – tomorrow’s challenges

Today’s customer experience goals depend on tightly aligned technical integration more than ever before, but it must be able
to handle trends going forward or risk becoming outdated.

Some of the high-level technical goals include: • Effective data “plumbing” for real-time web applications
and low latency
• Consolidated platforms that work together efficiently
• Act as a service layer that pushes data as close
• Simplified system architectures that are easy to manage
to the customer as possible

Five advantages of NoSQL databases

Ambitious customers with big ideas for how to use data have unleashed a new set of technology requirements for CIOs
and technical leaders.

Here are five trends that play into the advantages of NoSQL databases for addressing the challenges of building software.

Trends Requirements

• Scaling to support thousands or even millions of users

Customer shift
• Meeting UX requirements with consistently high performance
continues online
• Maintaining availability 24 hours a day, 7 days a week

• Supporting many different applications with different data structures

The internet is
• Ensuring software is “always on” with no excuse for downtime
connecting everything
• Supporting continuous streams of data from the real-time web

• Storing customer-generated semi-structured and unstructured data

Big data is • Storing different types of data from different sources in the same infrastructure
getting bigger or even the same cluster
• Storing data generated by thousands or millions of customers and IoT devices

• Scaling on demand to support more customers, and store more data

Applications are • Operating fully managed applications on a global scale to support
moving to the cloud customers worldwide
• Minimizing infrastructure and operating costs, achieving a faster time to market

• Creating “offline-first” apps – network connection not required

The world has
• Synchronizing mobile/edge data with remote databases in the cloud
gone mobile
• Supporting multiple mobile platforms with a single backend

The above requirements are extensive and challenge even the best systems to do even more with less. Today’s extreme
requirements can be loosely grouped into two categories that impact two different levels of end users:

• Providing agile platforms for application developers to excel

• Supporting scalable system architectures that outperform others

WHITEPAPER 4
Develop with agility

To remain competitive, enterprises must innovate – and

now they have to do it faster than ever before. Developers
are under extraordinary pressure. Speed is critical, but so
is agility, since these applications evolve far more rapidly
than legacy applications. Relational databases have a
restricted, flat data structure and don’t respond well to
frequent changes in the data model. This often impedes the
needs of modern, agile projects, applications, and business
requirements.

Adaptable NoSQL schema requirements

A core principle of agile development is responding to
change. When the requirements change, the data model also
changes. This is a problem for relational databases because
the data model can be cumbersome to change. In order
to change the data model, developers have to modify the
schema, or in some organizations, formally request a “schema
change” from the database administrators. This slows down
or stops development, not only because it is a manual,
time-consuming process, but because it also impacts other
applications and services.

WHITEPAPER 5
Flexibility for faster development
By comparison, a NoSQL document database fully supports agile development, because it is schema-less and does not
statically define how the data must be modeled. Instead, it defers to the applications and services, and thus to the developers
as to how data should be modeled. With NoSQL, the data model is defined by the application model. Applications and services
model data as objects.

Simplicity for easier development

Applications and services model data as objects (e.g., employee profile), multi-valued data as arrays (e.g., roles), and related
data as nested objects or arrays (e.g., manager relation). However, relational databases model data as tables of rows and
columns – related data as rows within different tables, multi-valued data as rows within the same table. One problem with
relational data modeling is that data is read and written by disassembling, or “shredding,” (like a paper shredder turns a single
document into multiple smaller strips) and reassembling objects. This is the object-relational “impedance mismatch.”

The workaround is object-relational mapping (ORM) frameworks, which can be effective in simple scenarios, but can become
problematic in more complex situations.

Consider an application for managing resumes. It interacts with resumes as an object of user objects. It contains an array for
skills and a collection for positions. However, writing a resume to a relational database requires the application to “shred” the
user object.

WHITEPAPER 6
Storing this resume would require the application to insert six rows into three tables, as illustrated in Figure 3.

However, reading this profile would require the application to read six rows from three tables, as illustrated in Figure 4.

In contrast, a document-oriented NoSQL database reads and writes data formatted in JSON – which is the de facto
standard for consuming and producing data for web, mobile, and IoT applications. It not only can eliminate the
object-relational impedance mismatch, it can eliminate the overhead of ORM frameworks and simplifies application
development because objects are read and written without “shredding” them – i.e., a single object can be read or
written as a single document, as illustrated in Figure 5.

WHITEPAPER 7
 Querying using SQL
Application developers that are used to querying with
SQL can continue to use the same language in NoSQL
platforms but operate against the JSON data that is
stored. For example, Couchbase provides a SQL-based
Grouping documents to ease access query standard known as SQL++ (sometimes known as

Unlike using predefined sets of schemas to differentiate N1QL) that returns results in JSON with sets of rows and

tables from one another, NoSQL databases have a subdocument components where appropriate. This is in

concept, such as buckets, that serve as a general holding contrast to the vast majority of other NoSQL databases

area for all documents. A database can have many (like MongoDB™) that don’t use SQL and require

logical, named, buckets for various purposes. The name is developers to climb a new language learning curve.

provided while connecting or requesting data and allows Standard statements are supported including SELECT
applications to have their own area in the system. … FROM … WHERE syntax. SQL++ also supports

Within those buckets are additional hierarchical logical aggregation, sorting, and joins (GROUP BY … SORT BY …

groupings that can be restricted to particular users LEFT OUTER/INNER JOIN). Querying collections, scopes,

or roles. These are called collections and/or scopes, and even nested arrays is supported. Query performance

allowing subsets of documents in a bucket to be named. can be improved with composite, partial, covering indexes,

Because this flexibility helps segregate data of one user and more.

or application from another, the developer does not have There are minimal changes for SQL experts to move to
to build their own security and reliability code, but can SQL++ where desired, with many basic queries working
instead let the underlying database do it. out of the box.

SQL SQL++

SELECT p.FirstName + ' ' + p.LastName AS SELECT p.FirstName || ' ' || p.LastName AS
Name, d.City Name, d.City
FROM AdventureWorks2016.Person.Person AS p FROM AdventureWorks2016.Person.Person AS p
INNER JOIN AdventureWorks2016. INNER JOIN AdventureWorks2016.
HumanResources.Employee e HumanResources.Employee e
ON p.BusinessEntityID = e.BusinessEntityID ON p.BusinessEntityID = e.BusinessEntityID
INNER JOIN INNER JOIN
(SELECT bea.BusinessEntityID, a.City (SELECT bea.BusinessEntityID, a.City
FROM AdventureWorks2016.Person.Address FROM AdventureWorks2016.Person.Address
AS a AS a
INNER JOIN AdventureWorks2016.Person. INNER JOIN AdventureWorks2016.Person.
BusinessEntityAddress AS bea BusinessEntityAddress AS bea
ON a.AddressID = bea.AddressID) AS d ON a.AddressID = bea.AddressID) AS d
ON p.BusinessEntityID = d.BusinessEntityID ON p.BusinessEntityID = d.BusinessEntityID
ORDER BY p.LastName, p.FirstName; ORDER BY p.LastName, p.FirstName;

WHITEPAPER 8
What about ACID transactions in NoSQL?
NoSQL databases also operate as operational systems with large numbers of transactions. When you flatten out a business
entity into multiple separate tables, you require a transaction for almost every update. With NoSQL databases, you don’t need
to flatten out the entity, but can usually contain it in a single document. Updates to a single document are atomic and don’t
require a transaction.

However, there may be updates that span multiple documents and require a check to ensure “all or nothing” of the transaction
occurs. For instance, a transfer of credits from one user’s account to another.

This is why NoSQL databases like Couchbase support transactions.

Java example:

Transactions transactions = Transactions.create(cluster,

TransactionConfigBuilder.create()

.durabilityLevel(TransactionDurabilityLevel.PERSIST_TO_MAJORITY)
.logOnFailure(true, Event.Severity.WARN)
.build());

TransactionResult result = transactions.run((ctx) -> {

// Inserting a doc:
ctx.insert(collection, “doc-a”, JsonObject.create());

// Getting documents:
// Use ctx.getOptional if the document may or may not exist
Optional<TransactionGetResult> docOpt =
ctx.getOptional(collection, “doc-a”);

// Use ctx.get if the document should exist, and the transaction

// will fail if it does not
TransactionGetResult docA = ctx.get(collection, “doc-a”);

// Replacing a doc:
TransactionGetResult docB = ctx.get(collection, “doc-b”);
JsonObject content = docB.contentAs(JsonObject.class);
content.put(“transactions”, “are awesome”);
ctx.replace(docB, content);

// Removing a doc:
TransactionGetResult docC = ctx.get(collection, “doc-c”);
ctx.remove(docC);

ctx.commit();
});

WHITEPAPER 9
The combination of transactions and SQL greatly expand
the number of use cases where a NoSQL database can
be considered. In the past, the inability to join or handle
transactional operations meant that NoSQL databases
were only chosen for the highest volume and scale use
cases. But the option of using SQL and transactions means
that NoSQL databases can also be chosen for traditional
RDBMS cases that need more flexibility and power.

Additionally, by selecting a transactional NoSQL database,

many traditionally complex applications can be simplified
because there is no need for an ORM tool.

One data source –

multiple access methods
NoSQL databases operate as a primary content store,
meaning you enter the data in one application but can
access it multiple ways depending on the use case. For
example, developers can use direct API calls to access a
specific document using a key or through a SQL query
that returns multiple rows of data in a JSON response. This
is known as “multi-model.”

Other access methods are available depending on the

database, including full-text search systems that allow
natural language search requests. Requests can be made
for full or partial “fuzzy” matches, geographic ranges,
or wildcard searches. The response includes a JSON
document with lists of matching document IDs, contextual
information, and a relevancy score.

Full-text search systems are often separate from a

database but NoSQL databases like Couchbase include
them as part of the underlying system, allowing managers
to simplify the overall architecture.

Big data analytics are possible as well, using

complimentary subsystems that process larger volumes
of historical data. Using advanced indexing and query
capabilities, also based on SQL++, more advanced
analytics can be done in the same database without
needing a separate, external OLAP system.

Because the data is stored and indexed all within the one
database product, it allows developers to connect to one
system and pass through the relevant requests.

WHITEPAPER 10
Operate at any scale

Databases that support web, mobile, and IoT applications must be able to operate at any scale. While it is possible to scale a
relational database like Oracle (using, for example, Oracle RAC), doing so is typically complex, expensive, and not fully reliable.
With Oracle, for example, scaling out using RAC technology requires numerous components and creates a single point of
failure that jeopardizes availability.

By comparison, a NoSQL distributed database – designed with a scale-out architecture and no single point of failure – provides
compelling operational advantages.

Elasticity for performance at scale

Applications and services have to support an ever-increasing amount of users and data – hundreds to thousands to millions of
users, and gigabytes to terabytes of operational data. At the same time, they have to efficiently scale to maintain performance.

The database has to be able to scale reads, writes, and storage. This is a problem for relational databases that are limited
to scaling up – i.e., only being able to scale by adding more processors, memory, and storage to a single physical server
(sometimes known as “vertical scaling”). As a result, the ability to scale efficiently, and on demand, is a challenge. It becomes
increasingly expensive, because enterprises have to purchase bigger and bigger servers to accommodate more users and more
data. In addition, it can result in downtime if the database has to be taken offline to perform hardware upgrades.

A distributed NoSQL database runs on commodity hardware to scale out – i.e., add more resources simply by adding more
servers to a cluster (sometimes known as “horizontal scaling”). The ability to scale out enables enterprises to scale more
efficiently by (a) deploying no more hardware than is required to meet the current load; (b) applying less expensive hardware
and/or cloud infrastructure; and (c) scaling on demand and without downtime.

By distributing reads, writes, and storage across a cluster of nodes, NoSQL databases are able to operate at any scale.
Additionally, they are designed to be easy to configure, install, and manage both small and large clusters.

WHITEPAPER 11
Availability for always-on, global deployment
As more and more customer engagements take place
online via web and mobile apps, availability becomes a
major concern. These mission-critical applications have to
be available 24 hours a day, 7 days a week – no exceptions.
Delivering 24x7 availability is a challenge for relational
databases that are deployed to a single physical server or
that rely on clustering with shared storage. If the single
server or shared storage fails, the database becomes
unavailable, applications stop, and customers disengage.

In contrast to relational technology, a distributed, NoSQL database partitions and distributes data to multiple database
instances with no shared resources. Couchbase goes a step further and does this automatically.

In addition, the data can be replicated to one or more instances for high availability (intercluster replication) and in different
geographic locations. While relational databases like Oracle require separate software for replication, for example, Oracle
Active Data Guard, NoSQL databases do not – it’s built in and it’s automatic. Couchbase’s cross data center replication (XDCR)
feature also provides this automatically.

In addition, automatic failover ensures that if a node fails, the database can continue to perform reads and writes by sending
the requests to a different node. Again, Couchbase will perform this failover recovery and rebalancing automatically.

Customer behavior often requires organizations to support multiple physical, online, and mobile channels in multiple regions
and often multiple countries. While deploying a database to multiple data centers increases availability and helps with disaster
recovery, it also has the benefit of increasing performance too. All reads and writes can be executed on the nearest data
center, thereby reducing latency.

Ensuring global availability is difficult for relational databases where separate add-ons are required – which increase
complexity – or where replication between multiple data centers can only be used for failover, because only one data center is
active at a time. Oracle, for example, requires Oracle GoldenGate. When replicating between data centers, applications built on
relational databases can experience performance degradation or find that the data centers are severely out of sync.

WHITEPAPER 12
A distributed, NoSQL database includes built-in replication between data centers – no separate software is required.
In addition, some include bidirectional replication enabling full active-active deployments to multiple data centers. This
enables the database to be deployed in multiple countries or regions while providing local data access to local applications
and their users.

Deploying to multiple data centers not only improves performance, but enables immediate failover via hardware routers.
Applications don’t have to wait for the database to discover the failure and perform its own failover.

NoSQL – Replication between

data centers is fully built-in,
and can be bidirectional

NoSQL is a better fit for large-scale requirements

As enterprises shift to cloud, mobile, social media, and big data technologies, developers, architects, and operations
teams have to build and maintain web, mobile, and IoT applications faster, and at a greater scale. NoSQL is increasingly
the preferred database technology to power today’s web, mobile, and IoT applications.

Hundreds of Global 2000 enterprises, along with tens of thousands of smaller businesses and
startups, have adopted NoSQL. For many, the use of NoSQL started with caching, proof of concept,
or a small application, then expanded to targeted mission-critical applications, and can become
the foundation for all application development. Today, the Couchbase NoSQL database serves
thousands of these types of customers.

With NoSQL, enterprises are better able to both develop with agility and operate
at any scale – and to deliver the performance and availability required to meet
the demands of businesses.

WHITEPAPER 13
 At Couchbase, we believe data is at the heart of the enterprise. We empower developers and

architects to build, deploy, and run their mission-critical applications. Couchbase delivers a high-

performance, flexible and scalable modern database that runs across the data center and any

cloud. Many of the world’s largest enterprises rely on Couchbase to power the core applications

their businesses depend on.

For more information, visit www.couchbase.com.

© 2022 Couchbase. All rights reserved.