KSII The 12th International Conference on Internet (ICONI) 2020.

Copyright ⓒ 2020 KSII

1

Comparative Study of Database Tools for

Android Application: A Bird’s Eye View
Hina Hussain*, Dr. Nawab Muhammad Faseeh Qureshi§, Dr. Qasim Ali*,
Dr. Isma Farah Siddiqui*¥
*
Department of Software Engineering, Mehran University of Engineering & Technology
Jamshoro, Pakistan
§
Department of Computer Education, Sungkyunkwan University, Seoul, South Korea
[e-mail: [email protected], [email protected]¥]
¥
Corresponding Author

Abstract

The database is the backbone of any software application. The prominent performance of the Database
is a significant factor in the development and efficient execution of android applications. Some
android apps need rapid input response, other processes much information, and without the support of
an efficient database management system, this is nearly incomprehensible. While talking about mobile
devices, then offline or local databases came into the picture. In this paper, the main focus is to
compare the performance of available offline tools/libraries which are SQLite, Room Persistence
Library, and ObjectBox in different scenarios. Firstly, perform CRUD operations on all three
databases and calculate time in seconds and compare the time of each operation. Secondly, calculate
the size of each database in MB. For this comparison, an android application has been developed to
test the performance and efficiency of each tool. The evaluation results show that the Object Box
library's performance is very well in terms of Create, Update, and Delete operations, and in Read
operation, whereas the SQLite is better than the other two. In terms of size Room Persistence Library
and SQLite, both are good than ObjectBox.

Keywords: Android, database, SQLite, Room Persistence Library, Object Box, CRUD, Performance

1. Introduction phones

The smartphone industry is consistently

developing and growing in terms of market size
as well as in models and suppliers. are expanding. Most Android applications
utilize some form of database to store particular
Right now, there are billions of Android cell
user's data and other types of information.
phones. Android is an open-source Linux based
operating system developed by Open Handset A database can be considered as the backbone
Alliance (OHA's), led by Google, and other of any application. The performance of the
companies. According to market research, database is significant in the construction of any
Android has a 72.92% market share.[1] The application. Android offers built-in database
number of accessible applications in the Google functionality in the form of SQLite but there are
Play Store from December 2009 to October a lot of libraries support database management.
2020 is 3.04 million.[2] As the market grows, Some of them are Room Persistence Library,
the hardware is becoming more powerful, and Object Box and Realm, and many more.
the possibilities of data processing from cell The main problem is to discover the current best
 2

solution on Android to store information on the solution for repeated and structured data for
device. Saving data in a database is an ideal
offline usage. However, most of the time, the increments in both classified and non-classified
developer community does not know which operations. N.Leavitt [6] discusses a brief
database library will be best according to their history of SQL and NoSQL databases close by
application requirements. the contrasts between them. He noticed that
The main goal of paper deals with the analysis SQL and NoSQL databases are complementary
of the performance of these three database since they are intended to overcome various
libraries in terms of most basic CRUD types of issues. N.Obradovic, A.Kelec and I.
operations, as well as in terms of memory size. Dujlovic [7] did a performance analysis of the
This comparison will help developers to Android SQLite database. Their study's primary
choose the best database library to increase the goal is to investigate the performance of the
performance of their apps. Section 2 focuses on SQLite database in different scenarios, such as
the literature review and research to deal with performing CRUD operations on encrypted or
databases and performance impacts. Section 3 unencrypted data. Further, they discuss
gives a brief overview of database libraries. simultaneous access to the database.
Section 4 describes the method and procedure.
This section also introduces an application,
called DBToolInspector, which performs 3. Overview of Selected Database
testing and performance measurement and tools
conducts results. Performance tests and their
results are discussed in section 5. 3.1 SQLite
SQLite is an open-source SQL database that
stores data to a text file on a device. Android
2. Literature Review
comes in with built-in SQLite database
Prior studies have been found that there are implementation. SQLite is the most used
many research work is available for database database engine in the world. [8]
comparison. A.Gupta, S.Tyagi, N.Panwar and S.
Sachdeva [3] did a crucial study and compared 3.2 Room Persistence Library
different types of NoSQL Databases based on Room is an ORM, Object Relational Mapping
functional and non-functional features.Four Library, which provides an abstraction layer
major classifications of the NoSQL database are over SQLite to allow easy database access. It is
graph databases, document stores, key-value a part of Android Jetpack and created By
stores, and wide column stores. Then for Google Developers.
additional investigation, one database is chosen
from everyone and is contrasted with their data 3.3 ObjectBox
model, CAP theorem, distributive properties,
ObjectBox is a super-fast database and
and other factors. A.Talha KABAKUŞ [4] did a
synchronization solution, built uniquely for
performance comparison of two widely used
Mobile and IoT devices [9], and created by
android databases, SQLite and Firebase. For this
GreenRobot.
comparison, he develops an android application
to execute common CRUD operation, and
experimental results indicate that SQLite has 4. Methods and Procedure
better performance than Firebase except in
delete operation. Benchmarking of these databases tools
S.Agarwal and K.Rajan [5] compare an SQL performance is very complicated because it
database (PostGIS) with a NoSQL database depends on different parameters ,such as device
(MongoDB) for spatial and aggregate queries. hardware ,OS version,file system and alse the
The tests they evaluate show that the NoSQL dataset.Therefore this paper is not focus on
database gives better execution by a normal external factors and datasets but focused on
factor of 10x-25x which increments database operations.These test cases have been
exponentially when the number of information conducted on real device Oppo A5s whose
 3

specifications are listed in Table 1.

Android API RAM CPU

Version Level
Android 27 2GB Octa-core
8.1.0 (Oreo) (4x2.3 GHz
Cortex-A53
& 4x1.8 GHz
8. Cortex-A53)

Fig. 2. DBToolInspectorApp interface

Table 1. Device Specifications

An android application named

DBToolInspector has developed to perform
CRUD operations on all three databases. A
5. Results and Discussions
Dataset used from google data sets which are By performing the CRUD operations, we have
“World Cities Population and Location” achieved the following results.
consists of 19 columns. DBToolInspector app
extracts data from the CSV file and performs 5.1 Create
CRUD operations on that datasets. We have
Fig.3 shows that while saving data the best
used different sizes of data sets such as 15000, result is ObjectBox. Furthermore, the worst
30000, 50000, 100000, and test cases are case is an SQLite database, and the Room also
executed 10x to calculate the average time in gives a reasonable result.
secs and size in MBS. The graph has been
shown to analyze time in each CRUD operation
and size in MB. Fig.1 shows the working flow
of the DBToolInspector app. Fig. 2 shows the
interface of the android application.

Fig. 3. Create Operation

5.2 Read
Fig.4 shows that when we are reading data, the
SQLite is incomparably faster than the other
DB systems. Room and ObjectBox take more
time than SQLite.
Fig. 1. Working flow of the app
 4

Fig. 4. Read Operation Fig. 6. Delete Operation

5.3 Update
When updating database information, again, 5.5 Size
ObjectBox has beaten SQLite, and SQLite has Comparing the presented systems, attention
the worst result, and Room Persistence Library should be paid to the memory used by the
has satisfactory results. The results can be databases. In this case, the worst is Object Box,
observed in Fig.5. whose database size is 40.48 MB. Room
Persistence Library and SQLite have achieved
almost the same results as shown in Fig. 7.

Fig. 5. Update Operation

Fig. 7. Size
5.4 Delete

For deleting data in Room, and SQLite has

performed very similar results. Again 6. Conclusions
ObjectBox is the best, but it should be noted that
the database cleanup surgery is rarely used. The The main goal of this paper is a comparison of
results can be observed in Fig.6. the database tool’s performance in terms of
CRUD and size. Evaluation results show that
ObjectBox performance is very good while
performing create, update, and delete
operations, but in terms of size, it has the worst
results. Moreover, when performing a read
operation, SQLite performance is far better than
the other two. Room Persistence Library
performs well in all operations. In terms of size,
 5

both SQLite and Room Persistence Library have [6] S.Agarwal and K.Rajan “Analyzing the
satisfactory results. performance of NoSQL vs. SQL databases
As the next step in research, our goal is to for Spatial and Aggregate queries,” Free
conduct more complex operations such as add and Open Source Software for Geospatial
more tables, perform joins and subqueries on (FOSS4G), Vol.17, Article 4,2017.
that data, and then compare results. However,
this does not mean that the result will be the [7] N.Leavitt, “Will NoSQL Databases Live
same for complex queries. Up to Their Promise? “Computer, 2010.

[8] N.Obradovic, A.Kelec and I. Dujlovic

“Performance analysis on Android SQLite
References database,” 18th International Symposium
INFOTEH- JAHORINA, 2019
[1] StatCounter Global Stats.
N.d. Mobile Operating System Market
Share Worldwide | Statcounter Global [9] Sqlite.org. n.d. SQLite Home Page.
Stats. [Online] Available at :< [online] Available
https://gs.statcounter.com/os-market- at:<https://www.sqlite.org/>.
share/mobile/worldwide>.
[2] Statista. N.d. Google Play Store: Number [10] ObjectBox. 2020. Objectbox - Edge
of Apps | Statista. [online] Available at: Database for Mobile and Iot. [online]
[3] <https://www.statista.com/statistics/ Available at:<https://objectbox.io/>.
266210/number-of-available-applications-
in-the-google-play- store/>.
[4] A.Gupta, S.Tyagi, N.Panwar and
S. Sachdeva, “NoSQL Databases: Critical
Analysis and Comparison,” Minjar Cloud
Services Private Limited, 2017.

[5] A.Talha KABAKUŞ “A Performance

Comparison of SQLite and Firebase
Databases from a Practical Perspective,”
DÜZCE ÜNİVERSİTESİ, MÜHENDİSLİK
FAKÜLTESİ, 2019.
6