CHAPTER NINE

NoSQL Polyglot Persistence

Ganesh Chandra Deka
Directorate General of Training, Ministry of Skill Development & Entrepreneurship, Government of India,
New Delhi, India

Contents
1. Introduction 358
1.1 Data Integration 358
2. Programming Paradigm 360
2.1 Polyglot Programming 363
3. Polyglot Persistence 365
4. Polyglot Persistence and NoSQL 366
4.1 NoSQL Databases Access Techniques 368
5. Big Data and Polyglot Persistence 370
5.1 Polyglot Processing 373
5.2 Polyglot-Persistent Databases 373
6. Polyglot Persistence in e-Commerce 374
6.1 REST and e-Commerce Polyglot Persistence 377
7. Polyglot Persistence in Healthcare 378
8. Research Trends in Polyglot Persistence 380
9. Conclusion 382
Glossary 383
References 384
Further Reading 389
About the Author 389

Abstract
Polyglot persistence facilitates use of most suitable database technology based on the
requirement of an application. There are lots of use cases as well as huge potential for
Polyglot Persistence in e-Commerce web portals, Search Engines, and Healthcare Infor-
mation Ecosystem applications. But, designing and implementation of an application in
a Polyglot environment is not a straightforward task, since adding more data storage
technologies increases complexity in programming. Further knowledge of multiple pro-
gramming languages, maintainability of the Polyglot application, and tool support for
Polyglot persistence are the challenges of Polyglot Persistence. This chapter discusses
the various aspects of Polyglot persistence with a focus on NoSQL database.

Advances in Computers, Volume 109 # 2018 Elsevier Inc. 357

ISSN 0065-2458 All rights reserved.
https://doi.org/10.1016/bs.adcom.2017.08.003
358 Ganesh Chandra Deka

1. INTRODUCTION
Polyglot persistence facilitates use of most suitable database technology
based on the requirement of an application. There are lots of use cases as well
as huge potential for Polyglot Persistence in e-Commerce Web portals, Sea-
rch Engines, and Healthcare Information Ecosystem applications. The survey
by research agency has revealed that the market value of retail e-Commerce
sales worldwide will be 4058 billion US dollars by 2020. Another survey by
451 Research commissioned by Oracle revealed that the market value of
DBaaS will reach $19.0 billion by 2020. Polyglot persistence application
development strategy will facilitate taking the benefits of Virtualization
Technology and Cloud Computing. This chapter discusses the Polyglot
Persistence as a data integration technology with a focus on NoSQL
database.

1.1 Data Integration

Eighty percent of data generated by various organizations as well as by indi-
viduals are unstructured. These data are in the form of audio recordings,
PDFs, and texts. Accesses to the data are usually in the form of a query,
reporting system/dashboard. The update of data can happen in bulk, real-
time, or near real-time.
The need for data integration emerges as the data generated by multiple
systems from complex environments are to be integrated. Data integration
facilitates uniform access to data available in multiple data sources such as
relational, nonrelational, and hierarchical database formats [1]. Data integra-
tion is a form of data preprocessing.
“An Average Fortune 1000 company has around 48 applications and
14 databases for a total of 62 potential systems which can be integrated” [2].
Data integration merges data from multiple data sources to provide:
• uniform query interface to all sources;
• access to queries; eventually updates too; and
• distributed over LAN, WAN even the internet.
Features of data integration are [3]:
• enables processing of data from a wide variety of sources;
• facilitates processing of unstructured data from social media, email, web
pages, etc.;
• syntactic and semantic checks to make sure that the data conform to busi-
ness rules and policies;
Application of NoSQL Database for Polyglot Persistence Application 359

• deduplication and removal of incorrectly/improperly formatted

data; and
• support for metadata.
The main technologies for data integration are (Fig. 1):
(a) extract, transform, load (ETL);
(b) enterprise application integration (EAI); and
(c) enterprise information integration (EII) aka data virtualization.
(a) The sources of data used in ETL include:
• Flat files
• Excel Sheet data
• Application data, such as CRM/ERP or Mainframe application data
Most difficult part of the ETL is the “Transform” component. Transform
process involves:
• Data cleansing/removal of duplicates
• Resolving issues relating to data inconsistency
• Apply rules to consistently convert data to appropriate form
(b) Features of EAI:
• Integrates applications and enterprise data sources for easy sharing
of data.
• Integrates applications and data sources within an enterprise to solve a
local problem.

Data source A

Data
warehouse
Data source B ETL

Data source C

Fig. 1 Data flow in ETL.

360 Ganesh Chandra Deka

Integration is done without significant changes in the applications and data

sources. EAI currently subsumed into service-oriented architecture soft-
ware stacks. EAI products are built in combination with the following
technologies:
• J2EE (Java 2 Platform Enterprise Edition);
• XML (eXtensible Markup Language) for enterprise-wide content and
data distribution;
• message queuing; and
• business process orchestration engine.
(c) EII aka data virtualization
• uses RDF (resource description framework) for information exchange;
• provides real-time Read and Write access;
• transform data for business analysis and data interchange; and
• manage data placement for performance, concurrency, and availability.
EAI and ETL are not competing for technologies. EAI can be source of
input or act as a service for ETL. An EAI-to-ETL interface could be used
to give an ETL product access to EAI application data. Such an interface
eliminates the need for ETL vendors to develop point-to-point (P2P)
adapters for these applications data sources.
EAI are focused on real-time processing, the EAI-to-ETL interface can
also act as a real-time event source for ETL applications requiring low-
latency data.
The ETL, EAI, and EII have their relative advantages and disadvantages,
but normally the combination is used. Table 1 summarizes the features of
ETL, EAI, and EII [4].
Polyglot persistence uses multiple database technology for storage and
retrieval of data. Hence, Polyglot persistence is basically a data integration
technique for an application from various data sources. Rest of the chapter
discusses about data integration using Polyglot persistence.

2. PROGRAMMING PARADIGM
Before discussing about the Polyglot persistence, the concept of
Polyglot programming is discussed in this section to build the foundation
of Polyglot persistence in the mind of the readers.
Following are the various programming paradigms:
(a) Functional programming
(b) Object-oriented programming
(c) Aspect-oriented programming
(d) Hybrid functional programming
Application of NoSQL Database for Polyglot Persistence Application 361

Table 1 Features of ETL, EAI, and EII

EIA ETL EII
• Integrates applications • Backbone technology • Leverages extended and
by using standard for data consolidation enhanced data
interfaces • In ETL, detailed warehousing
• Employs real-time profiling and analysis of environment to address
operational business the data sources and specific business needs
transaction processing their relationships to the • Similar to ETL
through remote targets are required processes, a detailed
procedure call (RPC), • Utilized for creating profiling and analysis
remote method physical data of the data sources and
invocation (RMI), warehouses their relationships to
common object request • ETL tools can the targets are required
broker architecture consolidate data into
(CORBA), or standard EAI targets and web
messaging services
infrastructures
• Can be used for
transportation of data
between applications
and route real-time
event data to other data
integration applications
such as an ETL process

(a) Functional programming is the oldest programming paradigm.

Functional programming also known as applicative programming and
value-oriented programming. Basically, functional programming is a
declarative type of programming. In functional programming, every-
thing is a function returning values instead of modifying data. Examples
are Lisp (historically, LIPS stands for List Processing is a programming
language for Artificial Intelligence), ML (meta language), Haskell,
Scala, F#, Erlang, etc. Haskell eliminate the side effects on variables,
making it easier for writing thread-safe code.
Functional programming languages are good for concurrency, but poor
in developing user interfaces (UIs). This problem can be overcome by a
Polyglot environment. The part of the software that needs to be highly
concurrent can be written using a functional programming language,
while the UIs part can be developed by using suitable general purpose
languages [5].
(b) Object-oriented programming is a way of programming that is
dominated by creating object (new data types) to solve a problem.
362 Ganesh Chandra Deka

An object contains both data and methods that manipulate data. For
example in C# everything is an object. Similar example of object-
oriented programming language includes Objective-C, Java, C++,
Python, and Smalltalk.
Out of all the object-oriented programming language Java and C++
are the mostly used. However, both of them are related with certain
advantage and disadvantages.
The following features available in C++ are not available in Java:
• Pointer
• Multiple inheritance
• Operator overloading
• GOTO statement
• Structure and union data structure
(c) Aspect-oriented programming (AOP) is similar to event handling.
The “events” are defined outside the code spanning over many
unrelated modules. In AOP crosscutting (spanning multiple software
modules) functionalities are implemented in Aspects instead of fusing
them into core modules. Aspects are an additional unit of modularity,
which can be reused for reducing code. An “Aspect weaver” takes the
aspects and the core modules and composes the final system.
The AOP programming language AspectJ provides additional keywords
and constructs to write Aspects. Aspects operate with reference to features
in the standard Java code. An aspect can have ordinary Java code, i.e.
• Member variables and methods
• Can implement interfaces
• There is also aspect inheritance
The AspectJ compiler produces standard Java byte code executable on any
Java Virtual Machine. The runtime library “aspectjrt.jar” is required to run
any AspectJ program. AspectJ and C# also derive a set of features from C and
C++ except pointers.
Another important feature of AspectJ (using the JDBC library) and C#
(using the ADO.NET library) is Database interaction. However, C# has
more features for supporting functional programming than AspectJ [6].
C# is suitable for mixing OOP and declarative style using Microsoft
LINQ (language-integrated query) services, especially for querying database
and XML files [6] (Table 2).
(d) F# and Scala are Hybrid Functional programming languages, i.e., a
mixture of object-oriented programming and functional programming
[7]. The object-oriented programming reduces the gap between the
Application of NoSQL Database for Polyglot Persistence Application 363

Table 2 Features of Programming Paradigms

Functional Object-Oriented Aspect-Oriented
Features Programming Programming Programming
Fundamental Function Object Aspects
working block
Variable Not used Used Used
Working level Function Object Code

problem domain and the implementation in software, while functional

programming brings robustness to computing power for concurrent
applications [8].
The term Polyglot (knowing/using several languages) programming
refers to developing software/computer applications using a mix of pro-
gramming languages. The idea behind this technique is to take advan-
tage of different languages suitable for tackling different problems.
Web-based applications are polyglot in nature using multiple programming
languages like:
• JavaScript, Flash, and HTML;
• UI such as Java, C#, and Ruby in the Middle Tier; and
• SQL/NoSQL as Database in a single program.
There are more than 256 programming languages, as listed at [9].
Some of the programming languages such as HTML are called Non-
computational languages. Often a Computational programming language
is embedded in the Noncomputational language specifically in internet-
based applications. According to Ref. [10], “The times of writing an appli-
cation in a single general purpose language are over.”
Modularity paradigms in application development are defined as poly-
paradigm programming (PPP).

2.1 Polyglot Programming

Polyglot programming is using several programming languages in software/
application [11]. Table 3 shows some of the use cases of ployglot
programming [12].
The degree of polyglotism determines the use of Polyglot programming.
The degree of Polyglot programming is determined by:
• integration of the programming language used;
• organization of code;
364 Ganesh Chandra Deka

Table 3 Polyglot Programming Examples

Sl. No. Name of Programming Languages Polyglot Use Case/Application
1 Java, XML, JavaScript, and HTML JSP (JavaServer Pages)
2 SQL and HTML PHP (Hypertext Preprocessor) pages
3 Ruby, JavaScript, and F#, in Microsoft Corporation
addition to Python, C#, C ++, and
Visual Basic
4 Python, JavaScript, Java, Ruby, Sun Microsystems
Groovy, and Scala
5 Ruby, Groovy, and Python JSR 223 Servlet (Java Specification
Request)
6 Java libraries in other languages Jython for Python, JRuby for Ruby
Java virtual machine (JVM) is created
to run on as many platforms as
possible to make Java platform
independent
Note: F# is an Open Source Cross Platform programming language [http://fsharp.org/].

Table 4 Applications Using Multiple Programming Languages [12]

Sl. No. Application Programming Languages Used
1 Civilization 4 (a computer game) C ++ and Python
2 World of Warcraft (a computer game) C ++ and Lua
3 Web application Cascading style sheets (CSS),
JavaScript, and HTML

• processes of program execution of programming language used; and

• data to be processed/manipulated.
The architecture where Polyglot programming can be utilized is [12]:
• Service-oriented architecture (SOA);
• Managed runtimes; and
• C integration;
The disadvantage of Polyglot programming includes (Table 4):
• knowledge required to use different languages;
• maintainability of different languages; and
• tools supported by different languages.
Hence, it is observed from the above discussion that a limited number of
languages are possible to use in an application. Google restricts the use of
Application of NoSQL Database for Polyglot Persistence Application 365

programming languages to JavaScript, Python, Java, and C ++ [13]. Use

case of Polyglot Programming is a web application’s UI implemented in
an OOP language Ruby, while services are implemented in a functional
programming language like Erlang.

3. POLYGLOT PERSISTENCE
Polyglot persistence is a hybrid approach enabling usage of multiple
databases in a single application/software. A Software that is capable of using
more than one type of data storage is referred to as Polyglot-persistent
software.
It is beneficial to use a variety of data models suitable for different parts of
huge software. For instance, use relational database to store/handle struc-
tured, tabular data; a document database such as MongoDB for unstructured,
object-like data; a key–value store such as Riak for a hash table; and a graph
database like Neo4j for highly linked referential data [14].
Fig. 2 shows a typical Polyglot persistence use case.
An example of Polyglot-persistent software is e-Commerce applications.
An e-Commerce application deals with multiple types of data, such as shop-
ping cart, inventory, customers’ orders, etc. To store all these data of an
e-Commerce application in one database will require lots of data conversion
work to make the data compatible originating from various sources [15].
Furthermore, implementing business transactions that maintain consistency
across multiple services is another challenge. Since, transactions are short-
lived session involving comparatively less data, hence NoSQL databases

Session data,
shopping carts
Key/Value

Product catalog,
user accounts,
MongoDB domain objects
e-Commerce application

Payment
RDBMS systems,
reporting

Graph
Social data,
recommendations

Fig. 2 Polyglot persistence.

366 Ganesh Chandra Deka

are suitable. However, once the transaction is over the buyer Account must
be Debited and sellers Account must be created. Here ACID transaction is
used for which SQL database (RDBMS) are the best fit [16].
A similar example of Polyglot persistence is microservices-based applica-
tions. These applications often use a mixture of SQL and NoSQL databases [16].
Polyglot-persistent application deployed on-premise relying on local
database or using multiple databases in the cloud using Database-as-a-
Service (DBaaS) offers of cloud service providers. However, implementing
Polyglot persistence in a distributed data management environment is a big
challenge. Furthermore, since NoSQL does not enforce data integrity,
hence data integration becomes the application’s concern for using NoSQL
database for e-Commerce applications. However, some of NoSQL data-
bases are ACID compliant at the cost of price and complexity. Hence DBaaS
will be the affordable and convenient option for developing applications/
software using multiple databases.

4. POLYGLOT PERSISTENCE AND NoSQL

NoSQL databases are designed to handle exponentially growing data
intensive applications. Most of the NoSQL databases are open source, how-
ever proprietary option is also available with added advantages/features.
NoSQL provide poor transactional integrity, flexible indexing, and que-
rying. The Polyglot persistence uses a mediator such as a REST API to issue
queries for NoSQL. The mediator acts as a broker between Application and
backend Database. This section discusses about the features of mostly used
NoSQL database with a focus on Polyglot persistence for interoperability.
Since the language used to develop the source code of the NoSQL deter-
mines lots of its underpinning properties, let us start with the programming
language used for developing the NoSQL.
As shown in Table 5, the Document NoSQL database is developed by
using both C++ and Erlang by the developer from various software vendors.
However, the Key–value NoSQL database is found to be more generic
since the C, C++, Java, and Erlang programming language were used to
write the source code of the database.
The most interesting observation is that, the majority of NoSQL database
was developed by using compatible programming languages (C is the basis
for C++, C#, and Java). Although it is not possible to run the C on Java
platform, there are facilities for invoking code written in C.
Application of NoSQL Database for Polyglot Persistence Application 367

Table 5 Programming Languages Used in 19 NoSQL and Search Engines

Sl. No. Name of NoSQL Category Programming Language Used
1 Bigtable Column C
2 MySQL MySQL based
3 Redis Key–value
Total 3
1 ClearDB MySQL based C#
2 Xeround MySQL based C#/C ++
Total 2
1 Hypertable Key–value C ++
2 Xeround MySQL based
3 MongoDB Document
4 MarkLogic Search Engine
Total 4
1 CouchDB Document Erlang
2 Amazon DynamoDB Key–value
3 Dynomite Key–value
4 SimpleDB Document
Total 4
1 Voldemort Key–value Java
2 Cassandra Column
3 HBase Column
4 Infinispan Data grid cloud
5 Elasticsearch Search Engine
6 PNUTS Column Java/JVM
Total 6

Java was used to develop the source code of four NoSQL database, one
Search Engine and one data grid cloud. Java runs on any platform by running
a JVM. Groovy, JavaScript, Python, Ruby, Perl, and PHP can all run on Java
platforms.
368 Ganesh Chandra Deka

It is observed from Table 5 that, two NoSQL databases, i.e., ClearDB

and Xeround, were developed using C#, a generic programming language
having high interoperability with C/C++.
Erlang is used to develop four NoSQL databases. Erlang has a very
interesting built-in primitive for message passing between processes, with-
out worrying about race conditions and deadlocks because of the functional
nature of this programming language.

4.1 NoSQL Databases Access Techniques

The following are some of NoSQL databases and the application for which
they were specifically designed (Table 6).
The NoSQL databases use various techniques to access data. Some
NoSQL databases support SQL but most NoSQL databases uses program-
matic querying through their unique custom APIs. For instance, Post-
greSQL RDBMS provides XML, JSON, and Key–Value functionalities.
There are Data Mappers which store whole object graphs into
Key–Value stores, since Key–value stores often allows limited querying.
Key–value store is the main technology beneath the relational database
management systems (RDBMS). In Key–value NoSQL database challenge
of having fast indices is solved using the B-tree approach [17].
Document stores generally allow more complex queries. Many
NoSQL databases have plug-in to Hadoop for Map/Reduce style of
“querying.”
Apache HBase allows random, real-time read/write access to big data.
Column families are used for optimization and partitioning purposes.

Table 6 NoSQL Databases Applications

Name of NoSQL Type of Data/Application Suitable for
MongoDB Product catalog
Redis Shopping cart
Amazon DynamoDB Social profile information
Neo4j Social graph
HBase Inbox and public feed messages
MySQL Payment and account information
Cassandra Audit and activity log, a query-based model
Application of NoSQL Database for Polyglot Persistence Application 369

MarkLogic uses single C++ core to solve the data management problem
across many data types from Text to RDF and XML to Binary. Neo4j Doc
Manager enables Polyglot persistence for Neo4j and MongoDB [18].
The D2R declarative language describes mappings between relational
database schemas and the OWL (web ontology language). The mappings
are used by the D2R processor to export data from a relational database into
RDF [19].
Another use case of Polyglot persistence is Force.com integrating and
optimizing varieties of data persistence technologies to deliver transparent
Polyglot persistence. With Force.com, coding a single API is sufficient
for a given situation irrespective of the degree of complexity of the prob-
lem/application. Force.com has both RESTful API and web services
(SOAP-based) APIs as well as Salesforce Object Query Language (SOQL)
for constructing powerful database queries for Polyglot persistence [20].
Kundera is a JPA 2.1 ( Java Persistence API) compliant object data store
mapping library for NoSQL databases. Kundera currently supports Cassan-
dra, HBase, MongoDBand MySQL (a relational database) [21].
Amazon DynamoDB DBaaS supports both document and Key–value
NoSQL data store models for easy development of web, Gaming, and
IoT (Internet of Things) in particular.
In an identical Polyglot persistence scenario, a MongoDB (document
database) can be used as the principal storage (master database) and Redis
(an in-memory database) as a caching layer to accelerate the writes.
Similarly, LAMP (Linux, Apache (open source web server software),
MySQL (also MongoDB), and Perl/PHP/Python) an open source solution
stack of software that enables rapid development of dynamic websites and
web applications [22,23]. Once all the components are installed in a server,
the final requirement is Installing Dependencies depending upon the pro-
gramming language and OS (operating system) used. For example,
mod_wsgi to be installed for Apache and Python to work together, sim-
ilarly mysqliclass is used as interface between MySQL and PHP [24].
The following are the advantages and disadvantages of Polyglot persis-
tence [25] (Table 7).
A Polyglot application stack might include Redis as a caching layer,
MongoDB for collecting logs, Postgres for metadata, and Elasticsearch for
indexing and search. This use case will require the technical knowledge of
all the three databases and one search engine associated with the system if
maintained by an organization [26]. The cost effective solution for such type
of the disadvantages of Polyglot persistence is migration to cloud base DBaaS.
370 Ganesh Chandra Deka

Table 7 Polyglot Persistence Advantages and Disadvantages

Advantages Disadvantages
Faster response Difficulties of integrating different databases
Easy to scale up application Expertise of different databases
Power of multiple databases Management of resources of various databases.
at the same time Another disadvantage is the fragmented
environments caused by Polyglot persistence by
running different databases side by side.
The fragmentation increases both the operational
and development complexity

Table 8 Traditional Data vs Big Data

Parameter Traditional Data Big Data
Size Gigabytes to Petabytes to Exabyte
Petabytes
Data source Centralized Distributed
Data type Structured Semistructured and unstructured as well as
high percentage are multistructured
Data model Stable Flat schemas
Data Known complex Limited interrelated complexities among the
interrelationship interrelationship data

5. BIG DATA AND POLYGLOT PERSISTENCE

Big data have hardly any structure at all. Big data are extracted from
the source in its raw form. The Big data are stored in a massively distributed
file system. According to Gartner survey, 73% organizations have either
invested or have plan to invest in big data in the coming next 2 years [27].
Table 8 shows the features of big data and conventional data.
According to [28], challenges of big data analytics are:
(a) Challenges of dealing with Volume, Velocity, and Variety of big data
• scalable big and fast data infrastructures;
• coping with diversity in the data management landscape; and
• end-to-end processing and understanding of data.
Following are the roles of NoSQL database for dealing with big data at rest as
well as in motion.
Application of NoSQL Database for Polyglot Persistence Application 371

• Analyze huge quantity of data in less time.

• Simplify use of big data.
• Optimum utilization of existing resources for handling of big data.
• Utilize existing skills with value addition.
The essential features of NoSQL for dealing with big data are in Table 9.
Some of the big data analytics use cases are:
• Real-time intelligence
• Data discovery
• Business reporting
Hence, Polyglot persistence is the best option for big data analytics (real time,
predictive, cognitive, and contextual) for creating a hybrid data processing
environment combining the functionalities of RDBMS and NoSQL database.
Hadoop Ecosystem is one of the use cases of polyglot persistence in big
data analytics. The user of Hadoop includes:
• Amazon/A9
• Facebook
• Google
• New York Times
• Veoh (an internet television company)
• Yahoo!

Table 9 NoSQL Databases and Big Data

Flexibility Appropriate
NoSQL Horizontal in Data Big Data
Type Scalability Variety Types Suitable for
Key–value High High Yes Stores and retrieves opaque
stores data items and blobs identified
by Key–value
Column High Moderate Partially Stores huge amount of data,
stores but only retrieve a subset of
these fields in queries
Document Variable High Likely Application needs to filter
stores (high) records based on nonkey fields
or update individual fields in
records
Graph Variable High Maybe Complex relationships
databases between entities
372 Ganesh Chandra Deka

Hadoop Common • Libraries and other modules

• Hadoop distributed file system written in Java for

HDFS providing redundant storage for massive amounts of data
on Community Hardware

• A framework for job scheduling and cluster resource

Hadoop YARN management

Hadoop
• Programming model for large-scale data processing
MapReduce

Fig. 3 Hadoop Ecosystem.

Fig. 3 shows the components of Hadoop Ecosystem which works in inte-

gration for big data processing.
Other tools of Hadoop Ecosystem:
• Hive-Developed at Facebook is a Data warehousing system for Hadoop,
Hive facilitating data summarization, Adhoc queries, and the analysis of
large datasets stored in Hadoop compatible file systems. Hive also pro-
vides a mechanism for querying of data using a SQL like query language
called HiveQL. HiveQL supports table partitioning, clustering, and
complex data types.
• Pig–Pig is a data-flow language for expressing Map/Reduce programs
for analyzing large HDFS distributed datasets. Pig provides relational
(SQL) operators such as JOIN, Group By, etc. Pig is also having easy
to plug in Java functions.
• Cascading pipe and filter processing model.
• HBase database model built on top of Hadoop.
• Flume designed for large-scale data movement.
The big data analytic application must has the capabilities to integrate and
manage lots of tools of Hadoop ecosystem. Amazon introduced Elastic
MapReduce (EMR) based on Hadoop. Amazon web services developed
the following big data application:
(i) Amazon DynamoDB NoSQL.
(ii) Amazon Redshift.
Oracle has developed an Oracle NoSQL database with a focus on big
data analytics. Hadoop and Oracle databases complement each other in
environments where massive servers collect data requiring sophisticated
analysis.
Application of NoSQL Database for Polyglot Persistence Application 373

5.1 Polyglot Processing

Polyglot processing refers to capability for processing any kind of Data,
Workload, and Workflow. Polyglot processing combines various data
processing engines over NoSQL.
Apache Storm is an open source Distributed real-time computation sys-
tem developed by BackType which was acquired by Twitter. The Storm
provides real-time computation, is horizontally scalable, Guarantees no data
loss and extremely robust and fault-tolerant. Apache Storm is an open
source distributed real-time computation system. Apache Storm is lan-
guage agnostic making easy to reliably process unbounded streams of data.
Following five characteristics make storm ideal for real-time data processing
workloads [29]:
(i) Fast: Capable of processing one million of 100 byte messages per second
per node.
(ii) Scalable: Run across a cluster of machines.
(iii) Fault-tolerant: When a worker dies, automatically restart them, in the
event a node dies, the worker will be restarted on another node.
(iv) Reliable: Guarantees that each unit of data (tuple) will be processed at
least once or exactly once. Messages are only replayed when there are
failures.
(v) Easy to operate: Once deployed, storm is easy to operate.
Apache Spark is a fast and general compute engine for large-scale data
processing. Apart from just “Map” and “Reduce,” Apache Spark defines a
large set of operations. Apache Spark supports Java, Scala, and Python. Apache
Spark can process data In-memory, but, Hadoop MapReduce persists back
to the disk after a Map or Reduce action. Apache Spark needs lots of mem-
ory for smooth performance. Apache Flink has unified runtime engine [30].
The Apache Lucene based open source search engine Elasticsearch is
used in addition to another NoSQL database [31] in use in many use cases.
MarkLogic, a multimodel database has an embedded Search Engine in addi-
tion to its regular features.

5.2 Polyglot-Persistent Databases

The research trend in database technology is to develop hybrid database.
This hybrid databases can be thought of as a borderless data center. Hybrid
databases will be having both the ACID (SQL) and BASE (NoSQL) prop-
erties. These databases are compatible with many SQL applications and pro-
viding the scalability of NoSQL applications.
374 Ganesh Chandra Deka

The MongoDB connector for BI (business intelligence) platform allows

use of MongoDB data for SQL-based BI [32] applications. CouchbaseN1QL
also extends SQL for JSON [33a,33b].
MarkLogic database achieves the goal of Polyglot persistence via the
technique of multimodel storage [34].
OrientDB is a multimodel open source NoSQL database that combines
the power of graphs NoSQL and flexibility of document database [35].
The term NewSQL is coined by 451 group analysts Matt Aslett represents
“A DBMS that delivers the scalability and flexibility promised by NoSQL
while retaining the support for SQL queries and/or ACID, or to improve
performance for appropriate workloads” [36]. Therefore, the NewSQL is
likely to be the alternative to NoSQL/SQL databases for new OLTP appli-
cations [37]. The examples of popular NewSQL are:
(i) VoltDB;
(ii) Clustrix; and
(iii) NuoDB are two commercial projects that are also classified as NewSQL.
NewSQL databases support the relational model and uses SQL as the query
language [38].
Another very important development in Polyglot processing is the con-
cept of hybrid transaction analytical processing (HTAP) architecture. HTAP
uses In-Memory technologies to run transactions and analytics processing on
the same database [39].

6. POLYGLOT PERSISTENCE IN E-COMMERCE

According to the survey conducted by “eMarketer,” retail
e-Commerce sales worldwide will be US$ 4058 billion by 2020 for products
or services ordered using the internet via any device [40]. Furthermore, the
survey forecasts that, the sales of retail e-Commerce will grow by 200% from
US$ 1915 billion in 2016 to US$ 4058 billion by 2020.
This section discusses the e-Commerce web application with a focus on
Polyglot persistence.
A Polyglot-persistent e-Commerce application is a hybrid system using
multiple databases as per the requirement of an e-Commerce application.
The benefits of using Polyglot persistence in e-Commerce application are:
• Productivity
• Maintainability
Using dynamic programming language such Ruby not only offers a higher
level of abstraction but also frees the developers from lots of repetition.
Application of NoSQL Database for Polyglot Persistence Application 375

Applications

Customer agent Price analytics/ Transactional

Session data Logs Product info operational
relationships data
analytics

Price
updates

Key/Value Columnar Document Graph Hadoop RDB

Fig. 4 Use case with various NoSQL database.

In this section, the scope of Polyglot persistence in e-Commerce will

be discussed.
Fig. 4 shows a typical e-Commerce application using various types of
unstructured data types. The figure shows some of the NoSQL database type
suitable for storage/manipulation of data/record generated by a particular
module/process of a typical e-Commerce application.
Here both SQL and NoSQL will be used by the e-Commerce applica-
tion. Since, financial transaction must follow ACID, hence the transactional
data must be stored in an RDBMS (SQL), whereas the other e-Commerce
data, such as session data, user logs can be easily managed by NoSQL.
Another important fact in the figure is the use of graph NoSQL database,
a relationship intensive database to store/handle Customer–Agent
relationships.
In this illustration, five varieties of databases including four NoSQL data-
base type and one relational RDBMS (SQL) database type will be suitable
for handling the data generated from an identical e-Commerce application.
In general, the following databases are suitable for various types of
e-Commerce functionality (Table 10).
However, using multiple type of database in an application is neither
technically nor financially viable. The thumb rule is use “as many as necessary,
as few as possible.” Currently, SQL server, Oracle, and MySQL are the three
376 Ganesh Chandra Deka

Table 10 Database in e-Commerce

Date Source Access Type Suitable Database Type
User sessions Fast access for reads and writes. Key–Value
Data need not be durable
Financial data Needs durable transaction RDBMS
updates. Tabular structure fits
the data
Point of sale Lots of writes, infrequent reads RDBMS (for write/reads
for analytics modest)
Key–Value or document (for
high write/reads) or column
if focus is on data analytics
Shopping cart High availability across Document/Key–Value
multiple locations.
Inconsistent writes to be
merged
Recommendations Traversing links between Graph for complex, if simple
friends, product purchases, and column
product/service ratings
Product Catalog Lots of read, infrequent writes Document NoSQL
Reporting SQL interfaces with reporting RDBMS, column NoSQL
tools
Analytics Analytics on large cluster Column
User activity/CSR High volume of write on Key–Value or document
logs, social media multiple nodes
analysis
Source: J. Serra, What is Polyglot Persistence? 2015, July 1 [Web log post]. Retrieved August 24, 2017,
from http://www.jamesserra.com/archive/2015/07/what-is-polyglot-persistence/.

most preferred database options for e-Commerce websites (all having JOSN
APIs to act as a mediator).
To conclude, a feasible e-Commerce solution may:
• Use of one RDBMS (SQL database) for durable financial transactions
• For various unstructured data uses
(i) Cloud-based NoSQL databases services DBaaS
(ii) Third party REST API for communication/integration between data-
bases and applications. However, lots of NoSQL have their inbuilt
REST APIs. The web service receives and sends requests through REST
and serializes data in a well-understood and portable format like JSON.
Application of NoSQL Database for Polyglot Persistence Application 377

6.1 REST and e-Commerce Polyglot Persistence

REST (representational state transfer)-based web services where Everything
is a Resource. For the REST-based applications such as Online shopping,
Search services (engines), and Dictionary services are all Resource. REST
ignores component implementation details, but focuses on:
• Roles of components.
• Their interactions and interpretation of data elements.
Web APIs can be used as HTTP services and can be subscribed by any client
application starting from desktop to mobiles.
Following are some of the use cases of REST APIs:
• Twitter: REST API and streaming API with JSON. The functionalities
of the REST API for Twitter include Sampling, Searching, and Filtering
capabilities.
• Amazon: has a “product advertising API” in for product search,
reviews, etc.
• eBay: Many APIs for searching, buying, and posting.
• Flickr: Comprehensive API set, free for noncommercial use with client
code in many languages.
A similar example of Polyglot persistence is SmartphoneApp that helps driver
to find a parking spot might involving the following:
• an inventory of parking lots;
• a mapping tool;
• a payment process; and
• a customer relationship management tool.
For the above-mentioned application, a Smartphone requires a Graph data-
base, a Geospatial database, and a Payment Gateway. The graph data model
is capable of integrating other data models [41] such as:
• schema and instance of relational data can be modeled with graphs;
• XML document is modeled as a tree and a tree is an acyclic, undirected
graph;
• document stores consist of nested Key–value pairs, hence can be
modeled as a set of trees; and
• Key–value stores as a set of nodes of the property graph.
Using MongoDB it is possible to build location aware applications.
MarkLogic enables Polyglot persistence with a multimodel database capable
of storing both the unstructured binary files such as .xslx, .pdf, and .jpg as
well as the structured XML data extracted from an Excel file [34].
Key–value stores are the most useful for the case of IoT, as the data have
no specific structure and come with various numbers and sizes. Data are
378 Ganesh Chandra Deka

treated as a set of bytes supplemented with a unique key. Riak® TS (Riak

time series) is a database suitable for IoT and time series data [42]. Another
example of IoT-specific database is DB4IoT (a database for IoT [43], http://
db4iot.com/), exclusively running on Intel® Xeon® processors.
There are lots of open sources time series database such as OpenTSDB
suitable for storage of IoT generated data [44]. Amazon DynamoDB and
VoltDB are the two more options [45].
Given a data stream coming from different social networks sites stored on
NoSQL database (Neo4J and Mongo) an option for retrieving the informa-
tion may be using the UnQL pivot query [46]. UnQL is based on SQL with
an extension to query NoSQL database data. UnQL can be used for query-
ing data stored in JavaScript object notation (JSON) format too.
The following databases support JSON in data representation:
• SQL Server 2016 and higher version [47]
• Oracle Database [1,48]
• Cassandra [49]
The OPENJSON rowset function (currently applies to SQL server 2016
and higher version and Azure SQL database) converts JSON text into a
set of rows and columns. OPENJSON can be used to run SQL queries
on JSON collections or to import JSON text into SQL server tables [50].
The above mentioned are few instances requiring databases to store the
varieties of data. The DBaaS offers a way to simplify the management of
multiple databases in the above-mentioned cases. Revenue from DBaaS
offerings is expected to climb at a CAGR of 46% from $2.849 billion in
2015 to $19.0 billion in 2020 [51].

7. POLYGLOT PERSISTENCE IN HEALTHCARE

eHealth is a broad term for healthcare practice supported by ICT.
Another terminology mHealth, the use of mobile phones to collect and
access health information. At Johns Hopkins University as of May 2015,
there were over 140 mHealth projects across the developing world (Key-
note presentation by Dr. Alain Labrique of Johns Hopkins University at the
Catholic Relief Services 2015 ICT4Development Conference, Chicago,
May 27, 2015) [52].
In healthcare domain, a variety of systems with different data types from
different locations is connected via network. Hence, a single Health Mon-
itoring system can generate/collect huge volume of semistructured data
Application of NoSQL Database for Polyglot Persistence Application 379

Fig. 5 Healthcare sector data sources.

such as XML, JSON, or unstructured mission-critical real-time information.

Data generated by health monitoring system also includes text, numerical
data, images, audio, video, sequences, time series data form IoT sensors,
social media data, multidimensional arrays, etc. Fig. 5 shows the source of
data in healthcare sector.
To extract meaningful information from these varieties of data, these data
need to be linked together making healthcare information ecosystem a com-
plex distributed system. Hence, building electronic health information sys-
tem (EHIS) using a single data model increases the complexity of application
development.
Polyglot persistence can be utilized for handling the heterogeneous data
(big data) associated with EHIS. The core idea is to store structured data in a
relational database, while semistructured or nonstructured data (big data) are
to be stored in NoSQL database. EHIS built on “Big Data” technologies
should be capable of serving the needs of clinicians in real time and in
one environment.
The big data analytic application such as Hadoop can be used for analysis
of the healthcare data generated by the various sources as shown in Fig. 5 for
planning and decision making in healthcare.
The following are the major collector of healthcare big data.
• Pharmacies
• Laboratories
380 Ganesh Chandra Deka

• Imaging centers (X-ray, CT scan)

• Emergency medical services (EMS)
• Hospital information systems
• Electronic medical records
• Blood banks
• Birth and death records
Big data generated by IoT will be valuable to healthcare. The sensors and
wearable can be used to collect health data on patients in their homes and
push all of that data into the Cloud. Healthcare institutions and care man-
agers, using sophisticated tools, will monitor this massive data stream and the
IoT to keep their patients health condition.
MemSQL in-memory database supports fast ingest and concurrent ana-
lytics for sensors, delivering instant actionable insight of data [53a] for var-
ious applications.
According to Das [53b], “By 2020, chronic conditions, such as cancer
and diabetes, are expected to be diagnosed in minutes using cognitive sys-
tems that provide real-time 3D images by identifying typical physiological
characteristics in the scans. By 2025, artificial intelligence (AI) systems are
expected to be implemented in 90% of the U.S. and 60% of the global hos-
pitals and insurance companies.” Similarly, a survey by Research and Mar-
kets argues that the Virtual reality healthcare systems market will be $2.5
billion by 2020 [54].

8. RESEARCH TRENDS IN POLYGLOT PERSISTENCE

Polyglot persistence is likely to benefit traditional databases by remov-
ing unstructured and semistructured data from general purpose relational
databases [36].
Neo4j has already developed “Docker compose” for deployment of
Polyglot persistence. The Docker Image allows defining multiple Dock-
erContainers and link them to a single YAML configuration file [55].
OpenShift PaaS from Red Hat is Multilanguage, Auto-Scaling, Self-
service, and Elastic. OpenShift supports MongoDB, PostgreSQL, and
MySQL and multilanguage support, i.e., Java, Perl, Python, PHP, and
Ruby [56].
Table 11 shows some interesting global figures of the various commod-
ities/services relating to this chapter by 2020.
e-Agriculture focuses on the enhancement of agricultural and rural
development. e-Agriculture involves the application of Smartphones, Cloud
Application of NoSQL Database for Polyglot Persistence Application 381

Table 11 IoT, DBaaS, and Healthcare Information System Market by 2020

Services All Figures in Billion
Cloud Shift US$ 216 [57]
DBaaS US$ 19.0 [36]
NoSQL US$ 4.2 [58]
Retail e-Commerce sale US$ 4058 [40]
IoT healthcare market US$ 163.24 [59]
IoT healthcare market by 2021 US$ 136 [60]
Healthcare big data analytics US$ 24.55 [59]
IoT market size US$ 3.7 [61]
Number of IoT devices 50 billion [62]
Data generated 44 zettabyte [63]
Devices connected to the network 25 billion devices [64]
Healthcare information system US$ 54 (market research store, [65])
IoT revenue 19 US$ Trillion [66]

Computing, IoT, and Big Data Analytics. The application of IoT in food
factories converts the machines to cyber-physical systems with embedded
intelligence and local data processing directly communicating with other
machines [67].
IoT devices will be generating huge volume of unstructured Time Series
Data. Pushed by geographically distributed devices in milliseconds/nano-
seconds deployed in healthcare, e-Agriculture and other areas. In healthcare,
the telemetric and wearable IoT devices will be gathering the real-time
information on health metrics for risk assessment for premium calculation
of insurance company and health practitioners. Processing these huge
amount of data in real-time is possible through cloud services which act
as concentrators to capture and store data.
Big data analytics tool will unlock value/insight hidden in Zettabyte
(1 Zettabyte ¼ 270 bytes) of real-time data generated IoT devices/sensors.
These data are to be stored in Low cost/Open source NoSQL database such
as a Key–value store or a Column family database. Graph NoSQL databases
using relation are the natural choice for managing the connected IoT
devices. Open source Column family database, like Cassandra or HBase
can be used for storage of IoT devices generated data [68].
382 Ganesh Chandra Deka

In compatibility, proprietary architectures, platforms, and standards are

leading to vendor lock-in for the adoption of IoT in e-Agriculture. There
are more than 115 protocols used by IoT devices for exchange of communi-
cation [69]. According to a McKinsey paper, interoperability will unlock more
than “US$ 4 trillion per year in potential economic impact for IoT use in 2025,
out of a total impact of US$ 11.1 trillion.” Interoperability is essential for cre-
ating 40% of the potential value to be generated by IoT in various settings [70].
The IoT generated data are communicated over IP (internet protocol) to
the edge servers/cloud for storage/processing. IP Version 6 (IPv6) will
eventually become the default protocol for IoT for connecting the devices
to the internet [52]. The networks must allow a variety of devices to connect
transparently enabling high performance computing and real-time video and
multimedia on demand [71]. For data interoperability, Polyglot persistence
can play a big role.
Database of Things (DoT), [72], capable of storing the data in a time
series, set of data from one device in relation to data of another device is
highly demanded in the age of IoT.
Blockchain is the technological innovation of Bitcoin for the public
transaction ledger in a decentralized network. Digitally recorded data are
packaged in Blocks for storage in a linear chain. Each block in the chain
contains cryptographically hashed Bitcoin transactions. Every Block con-
tains a hash of the previous block for creating a chain of blocks. In lockchain
distributed transaction database (ledger), all transactions are owned and
monitored by everyone and controlled by none, i.e., there is no central data-
base. Every node joins the network voluntarily and an “Administrator” of
the Blockchain. Blockchain database is not stored in any single location,
hence the records are public, easily verifiable, and accessible to anyone on
the internet. Blockchain technology will radically improve banking, supply
chain, health data, financial assets, automobiles, e-Governance, and other
transaction networks [73]. Microsoft has already started the Blockchain as
a Service (BaaS) for the customers [74].
Ethereum (https://www.ethereum.org) and other cryptocurrencies
[75] in distributed computing platform are powered by Blockchain.

9. CONCLUSION
The real world applications require the data integrity offered by an
RDBMS in addition to the benefits offered by NoSQL databases. None
of the database meets all the requirements of a web-based application. Hence
Application of NoSQL Database for Polyglot Persistence Application 383

Polyglot persistence is using a variety of data storage tools/technology for

managing multiform data according to different properties and requirements
will easy the situation to some extent.
Designing and implementation of a Polyglot system is not a straightfor-
ward task, since adding more data storage technologies increases complexity
in programming. Further knowledge of multiple programming language,
maintainability of the Polyglot application, and tool support for Polyglot
persistence are the challenges of Polyglot persistence. It is observed from var-
ious studies that the number of databases used should be kept as small as pos-
sible. Selection of the data store must be based on the requirements to
minimize the number of databases. The thumb rule for Polyglot persistence
is to use “as many as necessary, as few as possible.”
Some of the approaches to implement Polyglot solutions in an organi-
zation are:
• Multiple lanes
• Polyglot mapper
• Nested database
• Hadoop database.
The NewSQL databases deliver the scalability and flexibility promised by
NoSQL while retaining the support for SQL queries. The NewSQL data-
base is likely to be another option for developing applications requiring the
NoSQL–SQL dual characteristics. Examples of popular NewSQL database
are VoltDB. Two projects Clustrix and NuoDB, a commercial are also clas-
sified as NewSQL. Hadoop Ecosystem components Pig (Key–value),
GraphX (Graph), Drill (SQL), SparkSQL (SQL), and HBase (Column)
are the best example of Polyglot persistence use case.

GLOSSARY
Scripting Languages are used for smaller applications, but can also be used in large
applications.
XML an embedded configuration language widely used in Java and .NETworld. XML text
format encodes DOM (document-object models) which is a data structure for web pages.
SAX (simple API for XML) is an alternative “lightweight” way to process XML. A SAX
parser generates a stream of events as it parse the XML file.
Polyglot persistence refers continued or prolonged existence of something using several lan-
guages. The Polyglot-persistent application uses multiple data storage, i.e., RDBMS,
NoSQL, BLOB, file in a single (minimum) application that best suits the requirements
of an application. The parameter which plays a vital role in the development of Polyglot-
persistent application are Portability, Flexibility, and strong RDBMS integration.
Nonnetworked type of Polyglot programming where the different languages are in the same
file.
384 Ganesh Chandra Deka

JSON (JavaScript object notation) is a commonly used in web to describe the data being
exchanged between systems. JSON is a simple schema less language created to support
web pages. JSON object enters database as a text string which is parsed to discover its
fields and types.
YAML (Yet Another Markup Language) represents data in human-readable notation, almost
a superset of JSON.
NewSQL systems are relational databases designed to provide real-time OLTP (online trans-
action processing) and conventional SQL-based OLAP (online analytical processing) in
big data environments.
Bitcoin—A protocol that supports a decentralized, pseudoanonymous, peer-to-peer digital
currency. The blockchain is the fundamental data structure of the Bitcoin protocol.

REFERENCES
[1] C. Boylan, Data Integration (Database Integration). Retrieved January 23, 2017, from
http://www.informationbuilders.com/database-integration, n.d.
[2] K.K. Mohan, A.K. Verma, A. Srividya, R.K. Gedela, QoS Considerations for
Integration-Enterprise Service Bus, Communications in Dependability and Quality
Management—An International Journal 12 (1) (2009) 47–56. doi: UDC 004.421.2:
519.6, CDQM.
[3] TrustRadius, Data Integration Overview—What Is Data Integration Software?
Retrieved January 24, 2017, from, https://www.trustradius.com/data-integration,
2016. 11921 North MopacExpy, Suite 330, Austin, TX 78759.
[4] R. Pethuru, Cloud Enterprise Architecture, CRC Press, 2012. http://www.crcpress.
com/product/isbn/9781466502321.
[5] E. Letuchy, Facebook Chat, Retrieved January 21, 2017, from, http://www.facebook.
com/note.php?note_id¼14218138919&id¼9445547199&index¼1, 2008.
[6] Rabah, S., Li, J., Liu, M., & Lai, Y. (2010). Comparative Studies of 10 Programming
Languages within 10 Diverse Criteria—a Team 7 COMP6411-S10 Term Report.
arXiv preprint arXiv:1009.0305.
[7] D. Pountain, Functional Programming Comes of Age, 2006. BYTE.com (August
1994). Archived from the original on 2006-08-27. Retrieved January 24, 2017.
[8] Wikimedia Foundation Inc., Polyglot (Computing), Retrieved January 21, 2017, from:
https://en.wikipedia.org/wiki/Polyglot_(computing), 2016. Available under the Crea-
tive Commons Attribution-ShareAlike License.
[9] R. Diana, The Big List of 256 Programming Languages-Check Out a List of 256 Pro-
gramming Languages, From ABC to Z Shell, Retrieved January 24, 2017, from,
DZone/Java Zone, 2013. https://dzone.com/articles/https://dzone.com/articles/big-
list-256-programmingbig-list-256-programming.
[10] M. Wrangler, Polyglot Programming. Retrieved January 24, 2017, from http://
nealford.com/memeagora/2006/12/05/Polyglot_Programming, Thought Works, a
software company and a community of passionate, purpose-led individuals, n.d.
[11] B. Kullbach, A. Winter, P. Dahm, J. Ebert, in: Program Comprehension in Multi-
Language Systems. Reverse Engineering, 1998, Proceedings. Fifth Working Confer-
ence on, 135–143, 1998. Available from, http://ieeexplore.ieee.org/iel4/5867/
15624/00723183.pdf?tp¼&arnumber¼723183&isnumber¼15624.
[12] H. Fjeldberg, Polyglot Programming-A business perspective (Master’s thesis), Norwe-
gian University of Science and Technology, Department of Computer and Information
Science, Trondheim, Norway, 2008, Norwegian University of Science and
Application of NoSQL Database for Polyglot Persistence Application 385

Technology, Department of Computer and Information Science, Trondheim, Norway,

2008, pp. 1–75. Supervisor: HallvardTrætteberg, IDI.
[13] S. Yegge, Rhino on Rails, Retrieved January 21, 2017, from, http://steve-yegge.
blogspot.com/2007/06/rhino-on-rails.html, 2007.
[14] M. Neunh€ offer, July 7, in: Data Modeling With Multi-Model Databases—A Case Study
for Mixing Different Data Models Within the Same Data Store, 2015. Retrieved Jan-
uary 23, 2017, fromhttps://www.oreilly.com/ideas/data-modeling-with-multi-model-
databases.
[15] MapR Technologies, Inc, Why MapR, Retrieved January 22, 2017, fromhttps://www.
mapr.com/why-hadoop/why-mapr/enterprise-platform, 2016. One Platform for Big
Data Applications.
[16] C. Richardson, Microservices and the Problem of Distributed Data Management,
Retrieved January 23, 2017, from, DZone Integration Zone, 2016. https://dzone.
com/articles/event-driven-data-management-for-microservices-par.
[17] E. De Oliveira, Why a Key–Value Store NoSQL Database Is the Best Solution for
Future IoT Needs, Retrieved February 01, 2017, from, FairCom Corporation, 2016.
https://www.faircom.com/insights/key-value-store-nosql-database-best-solution-future-
iot-needs.
[18] W. Lyon, Neo4j Doc Manager: Polyglot Persistence for MongoDB& Neo4j, Retrieved
January 25, 2017, from, Neo Technology, Inc., 2015. https://neo4j.com/blog/neo4j-
doc-manager-polyglot-persistence-mongodb/.
[19] C. Bizer, D2R Map—Database to RDF Map—Chris Bizer. Retrieved January 25,
2017, from http://wifo5-03.informatik.uni-mannheim.de/bizer/d2rmap/D2Rmap.
htm, FU Berlin, n.d.
[20] Salesforce.com, Inc, Multi Tenant Architecture–Developer. Force.com, Retrieved
January 26, 2017, from, San Francisco, CA. https://developer.salesforce.com/page/
Multi_Tenant_Architecture, 2016.
[21] K. Prasad, Polyglot Persistence, Retrieved January 26, 2017, from, GitHub, Inc, 2016.
https://github.com/impetus-opensource/Kundera/wiki/Polyglot-Persistence.
[22] Wikimedia Foundation Inc., LAMP (Software Bundle), Retrieved March 08, 2017,
from, https://en.wikipedia.org/wiki/LAMP_(software_bundle), 2017.
[23] K. Payne, A Step by Step Guide to Install LAMP (Linux, Apache, MySQL, Python) on
Ubuntu, Retrieved March 08, 2017, from, http://blog.udacity.com/2015/03/step-by-
step-guide-install-lamp-linux-apache-mysql-python-ubuntu.html, 2015.
[24] The PHP Group, Themysqli Class, Retrieved March 08, 2017, from, http://php.net/
manual/en/class.mysqli.php, 2017.
[25] N. Dhandala, The Pros and Cons of Polyglot Persistence, Retrieved January 25, 2017,
from, http://opensourceforu.com/2015/08/the-pros-and-cons-of-polyglot-persistence/,
2015. Tags: e-Commerce, Enterprise, Network, NoSQL, Persistence.
[26] S. Pimentel, The Rise of the Multimodel Database, Retrieved March 26, 2017, from,
http://www.infoworld.com/article/2861579/database/the-rise-of-the-multimodel-
database.html, 2015.
[27] Stamford, Gartner Survey Reveals That 73 Percent of Organizations Have Invested
or Plan to Invest in Big Data in the Next Two Years, Retrieved February 26, 2017,
from, Gartner, Inc, 2014. http://www.gartner.com/newsroom/id/2848718.
[28] D. Abadi, R. Agrawal, A. Ailamaki, M. Balazinska, P.A. Bernstein, M.J. Carey,
S. Chaudhuri, et al., The Beckman Report on Database Research, Retrieved January
28, 2017, from, Beckman Center of the National Academies of Sciences & Engineering
Irvine, CA, USA, 2013. http://beckman.cs.wisc.edu/beckman-report2013.pdf.
[29] Apache Storm, Retrieved January 22, 2017, from http://hortonworks.com/apache/
storm/, n.d.
386 Ganesh Chandra Deka

[30] S. Edlich, N*SQL, Retrieved January 24, 2017, from, http://nosql-database.org, 2011.
Owned by: Prof. Dr. Stefan Edlich Stubenrauchstr.9c 14167 Berlin Gemany.
[31] A. Fowler, NoSQL For Dummies (India ed.), Wiley India Pvt. Ltd. A Wiley Brand,
New Delhi, 2015.
[32] MongoDB, Inc, MongoDB Connector for BI, Retrieved February 18, 2017, from,
https://www.mongodb.com/products/bi-connector, 2017.
[33] (a) ReactiveCouchbase, 2017. ReactiveCouchbase/repository. Retrieved August 24,
2017, from https://github.com/ReactiveCouchbase/repository/blob/master/snap-
shots/org/reactivecouchbase; (b) R. Zicari, New Gartner Magic Quadrant for Opera-
tional Database Management Systems. Interview with Nick Heudecker, Retrieved
February 18, 2017, from, http://www.odbms.org/blog/2016/11/new-gartner-
magic-quadrant-for-operational-database-management-systems-interview-with-nick-
heudecker/, 2016 ODBMS.org.
[34] D. Feldman, Avoiding the Franken-Beast: Polyglot Persistence Done Right.Retrieved
January 24, 2017, from http://www.marklogic.com/blog/polyglot-persistence-done-
right/, MarkLogic®/Agile Development, Data Management, n.d.
[35] OrientDB Ltd, Why Orientdb?Graph/Document Multi-Model Database. Retrieved
January 26, 2017, from http://orientdb.com/why-orientdb/, OrientDB Open Source
Project. n.d.
[36] 451 Research, NoSQL, NewSQL and Beyond: The Drivers and Use-Cases for Data-
base Alternatives, Retrieved January 23, 2017, from, http://www.bing.com/cr?
IG¼548773F8E30444D19B7B1B3FA576F38F&CID¼365A0BE5B3476EDE2F1A01
F6B2766FB1&rd¼1&h¼g47O96F_f6AXs9xLPkf6FUjtDHpwEH142OcqpiMER6w
&v¼1&r¼https%3a%2f%2f2.zoppoz.workers.dev%3a443%2fhttps%2f451research.com%2freport-long%3ficid%3d1389&p¼Dev
Ex, 2011. 5083.1, The 451 Group: Commercial Adoption of Open Source.
[37] M. Stonebraker, New SQL: An Alternative to NoSQL and Old SQL for New OLTP
Apps, Retrieved February 16, 2017, from, http://cacm.acm.org/blogs/blog-cacm/
109710-new-sql-an-alternative-to-nosql-and-old-sql-for-new-oltp-apps/fulltext, 2011.
[38] K. Grolinger, W.A. Higashino, A. Tiwari, M.A.M. Capretz, Data Management in Cloud
Environments: NoSQL and NewSQL Data Stores, Retrieved February 16, 2017, from,
http://journalofcloudcomputing.springeropen.com/articles/10.1186/2192-113X-2-22,
2013. Journal of Cloud Computing.
[39] M. Pezzini, D. Feinberg, N. Rayner, R. Edjlali, Hybrid Transaction/Analytical
Processing Will Foster Opportunities for Dramatic Business Innovation, Retrieved
February 15, 2017, from, Gartner, Inc., 2014. https://www.gartner.com/doc/
2657815/hybrid-transactionanalytical-processing-foster-opportunities.
[40] Statista Inc., Global retail e-commerce sales 2014-2020, Retrieved January 26, 2017,
from, https://www.statista.com/statistics/379046/worldwide-retail-e-commerce-sales/,
2014Worldwide; eMarketer; 2014 to 2016.
[41] A. Maccioni, O. Cassanon, Y. Luo, J. Castrejon-Castillo, G. Vargas-Solar NoXperanto:
Crowdsourced Polyglot Persistence. [Research Report] RRLIG-047, 2014, http://
basho.com/use-cases/iot-sensor-device-data/.
[42] Basho, IoT/Sensor/Device Data. Retrieved January 29, 2017, from http://basho.com/
use-cases/iot-sensor-device-data/, n.d.
[43] A Database for Internet of Things (IoT) Data, Retrieved January 29, 2017, from http://
db4iot.com/. n.d.
[44] The OpenTSDB Authors, OpenTSDB—A Distributed, Scalable Monitoring System,
Retrieved January 29, 2017, from, http://opentsdb.net, 2016. OpenTSDB 2.3.0.
[45] VoltDB, Inc, IoT Sensors, Retrieved January 29, 2017, from, https://www.voltdb.
com/solutions/verticals/iot-sensors, 2016.
Application of NoSQL Database for Polyglot Persistence Application 387

[46] Dataversity, UnQL: A Standardized Query Language for NoSQL Databases, Retrieved
January 22, 2017, from, http://www.dataversity.net/unql-a-standardized-query-
language-for-nosql-databases/, 2013. Dataversity Education, LLC.
[47] Microsoft Developer Network, JSON Data (SQL Server), Retrieved January 26, 2017,
from, https://msdn.microsoft.com/en-us/library/dn921897.aspx, 2016.
[48] XML DB Developer’s Guide, 39 JSON in Oracle Database. Retrieved January 26,
2017, from http://docs.oracle.com/database/121/ADXDB/json.htm#ADXDB6246,
Oracle, n.d.
[49] DataStax, Inc, Inserting JSON Data Into a Table, Retrieved January 26, 2017, from,
https://docs.datastax.com/en/cql/3.3/cql/cql_using/useInsertJSON.html, 2017.
[50] Microsoft, Convert JSON Data to Rows and Columns With OPENJSON (SQL
Server), Retrieved February 19, 2017, from, https://msdn.microsoft.com/en-us/
library/dn921879.aspx, 2017. SQL Server 2016 and later.
[51] Oracle, Database as a Service (DBaaS): Use Cases and Adoption Patterns A Report on
Research Commissioned by Oracle, Retrieved January 26, 2017, from, http://www.
oracle.com/us/products/database/451-report-dbaas-3038788.pdf, 2016. 451 Research,
LLC and/or its Affiliates.
[52] International Telecommunication Union, Harnessing the Internet of Things for Global
Development, Retrieved February 5, 2017, from, A Contribution to The UN Broad-
band Commission for Sustainable Development, 2016. https://www.itu.int/en/action/
broadband/Documents/Harnessing-IoT-Global-Development.pdf.
[53] (a) MemSQL Inc, Real-Time Solutions for IoT. Retrieved February 05, 2017, from
http://www.memsql.com/solutions/internet-of-things/, n.d.; (b) R. Das, 2016, April
26, Five Technologies that Will Disrupt Healthcare By 2020. Retrieved August 31,
2017, from https://www.forbes.com/sites/reenitadas/2016/03/30/top-5-technologies-
disrupting-healthcare-by-2020/#51b467e16826.
[54] D. Erg€ urel, Virtual Reality Healthcare Systems Market to Reach $2.5 Billion by 2020,
Retrieved March 05, 2017, from, https://haptic.al/healthcare-virtual-reality-market-
9d23a2cc9cd6#.ze5plsida, 2016. Haptical.
[55] W. Lyon, NoSQL Polyglot Persistence: Tools and Integrations With Neo4j, Retrieved
January 23, 2017, from, Neo Technology, Inc, 2016. https://neo4j.com/blog/nosql-
polyglot-persistence-tools-integrations/#why-polyglot-persistence.
[56] S. Gulati, Thinking Beyond RDBMS: Building Polyglot Persistence Java Applications.
Retrieved January 25, 2017, from http://events.linuxfoundation.org/sites/events/files/
slides/Thinking%20Beyond%20RDBMS%20–%20Building%20Polyglot%20Persistence
%20Java%20Applications.pdf, The Linux Foundation®, n.d.
[57] Gartner, Gartner Says by 2020, Retrieved January 28, 2017, from, STAMFORD,
Conn., 2016. http://www.gartner.com/newsroom/id/3384720.
[58] L. Person, NoSQL Market by Type (Key–Value Store, Document Databases, Col-
umn Based Stores and Graph Database) and Application (Data Storage, Metadata
Store, Cache Memory, Distributed Data Depository, e-Commerce, Mobile
Apps, Web Applications, Data Analytics and Social Networking)—Global
Opportunity Analysis and Industry Forecast, 2013–2020, Retrieved January 28,
2017, from, Allied Market Research, 2015. https://www.alliedmarketresearch.
com/NoSQL-market.
[59] Home, Healthcare Analytics Market Worth 24.55 Billion USD by 2021. Retrieved
January 28, 2017, from http://www.marketsandmarkets.com/PressReleases/healthcare-
data-analytics.asp, n.d.
[60] M. A. Malik, Internet of Things (IoT) Healthcare Market by Component (Implantable
Sensor Devices, Wearable Sensor Devices, System and Software), Application (Patient
388 Ganesh Chandra Deka

Monitoring, Clinical Operation and Workflow Optimization, Clinical Imaging, Fitness

and Wellness Measurement)—Global Opportunity Analysis and Industry Forecast,
2014–2021. Retrieved January 28, 2017, from, https://www.alliedmarketresearch.
com/iot-healthcare-market, n.d.
[61] L. Columbus, Roundup of Internet of Things Forecasts and Market Estimates, 2016,
Retrieved January 28, 2017, from, Forbes, 2016. http://www.forbes.com/sites/
louiscolumbus/2016/11/27/roundup-of-internet-of-things-forecasts-and-market-
estimates-2016/#289c2e8e4ba5.
[62] S.Z. Aeris, 50B IoT Devices Connected by 2020—Beyond the Hype and Into Reality,
Retrieved January 28, 2017, from, Syed Zaeem Hosain, Cto, Aeris, 2016. http://
www.rcrwireless.com/20160628/opinion/reality-check-50b-iot-devices-connected-
2020-beyond-hype-reality-tag10.
[63] A. Adshead, Data Set to Grow 10-Fold by 2020 as Internet of Things Takes Off,
Retrieved January 28, 2017, from, Antony Adshead, Storage Editor–TechTarget–
ComputerWeekly, 2014. http://www.computerweekly.com/news/2240217788/Data-
set-to-grow-10-fold-by-2020-as-internet-of-things-takes-off.
[64] A. Staff, Terabyte Terror: It Takes Special Databases to Lasso the Internet of Things,
Retrieved January 29, 2017, from, WIRED Media Group, 2016. https://arstechnica.
com/information-technology/2016/06/building-databases-for-the-internet-of-data-
spewing-things/.
[65] Market Research Store, Global Healthcare Information System Market Set for
Rapid Growth, To Reach Around USD 54.0 Billion By 2020, Retrieved January 29,
2017, from, http://www.marketresearchstore.com/news/global-healthcare-information-
system-market-190, 2015.
[66] MongoDB, Internet of Things, Retrieved February 01, 2017, from, MongoDB, Inc.,
2017. https://www.mongodb.com/use-cases/internet-of-things.
[67] G.M. Siddesh, G.C. Deka, K.G. Srinivasa, L.M. Patnaik, Cyber-Physical
Systems: A Computational Perspective, CRC Press, Taylor & Francis Group, Boca
Raton, 2016.
[68] CITO Research, Six Essential Skills for Mastering the Internet of Connected Things.
Retrieved January 29, 2017, from http://www.evolvedmedia.com/wp-content/uploads/
2015/05/CITO-Research_Neo-Technology_Six-Essential-Skills-for-Mastering-the-
Internet-of-Connected-Things_White-Paper_2014.pdf, Sponsored by Neo4j n.d.
[69] G. Schatz, The Complete List of Wireless IoT Network Protocols, Retrieved February
06, 2017, from, LINK LABS, 2016. https://www.link-labs.com/blog/complete-list-
iot-network-protocols.
[70] J. Manyika, M. Chui, P. Bisson, J. Woetzel, R. Dobbs, J. Bughin, D. Aharon, The
Internet of Things: Mapping The Value Beyond The Hype-Executive Summary,
Mckinsey Global Institute, Retrieved February 5, 2017, from, McKinsey & Company,
2015. file:///C:/Users/HP%20LAPTOP/Downloads/Unlocking_the_potential_of_
the_Internet_of_Things_Executive_summary.pdf.
[71] Cisco, Cisco VNI Mobile Forecast (2015–2020)-Cisco, Retrieved February 06,
2017, from, http://www.cisco.com/c/en/us/solutions/collateral/service-provider/
visual-networking-index-vni/mobile-white-paper-c11-520862.html, 2016.
[72] S. Samaraweera, Database of Things (DoT): The Future of Database. Has It’s Time
Come? Retrieved February 05, 2017, from, Virtusa Corporation, 2015. http://blog.
virtusapolaris.com/2015/10/database-of-things-dot-the-future-of-database-has-its-
time-come/.
[73] S. Norton, CIO Explainer: What Is Blockchain? Retrieved February 18, 2017, from,
Dow Jones & Company, Inc, 2016. http://blogs.Wsj.Com/cio/2016/02/02/cio-
explainer-what-is-blockchain/.
Application of NoSQL Database for Polyglot Persistence Application 389

[74] The Windows Club, Blockchain Technology & Microsoft’s Blockchain as a Service
Strategy, Retrieved February 19, 2017, from, http://www.thewindowsclub.com/
blockchain-microsoft-plans-develop-service, 2016.
[75] O. Ogundeji, Blockchain News, Retrieved February 18, 2017, from, https://
cointelegraph.com/tags/blockchain, 2017.

FURTHER READING
[76] D.G. Chandra, M.D. Borah, in: Cost benefit analysis of cloud computing in education,
In Computing, Communication and Applications (ICCCA), 2012 International Con-
ference on, IEEE, 2012, pp. 1–6.
[77] Deka, G. C. (2014). A survey of cloud database systems IT Professional, 16(2), 50–57.
https://doi.org/10.1109/MITP.2013.1, Survey of 15 popular NoSQL Databases.
[78] D.G. Chandra, BASE analysis of NoSQL database, Future Gener. Comput. Syst.
52 (2015) 13–21.
[79] G.C. Deka, Handbook of Research on Securing Cloud-Based Databases With Biomet-
ric Applications, IGI Global, 2014.
[80] G.C. Deka, P.K. Das, An Overview on the Virtualization Technology. Handbook of
Research on Cloud Infrastructures for Big Data Analytics, IGI Global, 2014,
pp. 289–321.
[81] G.S. Tomar, N.S. Chaudhari, R.S. Bhadoria, G.C. Deka, The Human Element of Big
Data: Issues, Analytics, and Performance, CRC Press, 2016.
[82] Deka, G. C., Siddesh, G. M., Srinivasa, K. G., &Patnaik, L. M. (2016). Emerging
Research Surrounding Power Consumption and Performance Issues in Utility Com-
puting. Information Science Reference.
[83] Deka, G. C., Zain, J. M., & Mahanti, P., ICT’s role in e-Governance in India and
Malaysia: a review, (2012) arXiv Preprint arXiv:1206.0681.

ABOUT THE AUTHOR

Ganesh Chandra Deka is currently Deputy
Director (Training) at Directorate General of
Training, Ministry of Skill Development and
Entrepreneurship, Government of India.
His research interests include ICT in
Rural Development, e-Governance, Cloud
Computing, Data Mining, NoSQL Data-
bases, and Vocational Education and Train-
ing. He has published more than 57 research
papers in various conferences, workshops, and
International Journals of repute including
IEEE & Elsevier. So far he has organized
08 IEEE International Conference as Technical Chair in India. He is the
390 Ganesh Chandra Deka

member of editorial board and reviewer for various Journals and International
conferences.
He is the co-author for four text books in fundamentals of computer sci-
ence. He has edited seven books (three IGI Global, USA, three CRC Press,
USA, one Springer) on Bigdata, NoSQL, and Cloud Computing in general as
of now.
He is Member of IEEE, the Institution of Electronics and Telecommunication
Engineers, India and Associate Member, the Institution of Engineers, India.