ABSTRACT
This study presents the Design and Performance Analysis of a Control System for Data Base System Management. Automatic tuning has been an elusive goal for database technology for a long time and is becoming a pressing issue for modern E-services. The research presented how relevant control-theoretical techniques can be used to solve typical problems in the database management system. One of the major difficulties in using formal methods to study such problems is the lack of accurate analytical models for complex database systems. The study presented a PID controller which combines Proportional action, Integral action and Derivative action in its operation making it the most complete controller available which provides a quick response, a control signal that tends to provide stability to the system and a minimum steady state error. The fuzzy Inference System (FIS) was presented to be able to design a database model suitable for this work. The research presented two models, one is the conventional model and the second is the optimized model. Some key performance indicators were examined in the conventional model involving two databases, database A and B. The key performance indicators includes; The database size, the database speed (flow rate), the database access rate and the database processing. The simulation results of the two models were compared in areas of data flow rate, data access rate, data flow error and also the simulation output rule behaviors. The result from the simulation proved the effectiveness of the optimized model which tends to provide stability to the system, a seamless flow of data and a minimum steady state error. This guarantees its ability to optimize the performance of the database. The developed optimized model was able to improve on the workload of the Database Management System (DBMS), Database stability is 60%, Optimized flow rate 50%, Database access rate 50% and the Database error rate is 30%.
TABLE OF CONTENTS
Title page i
Declaration ii
Certification iii
Dedication iv
Acknowledgements v
Table of Contents vi
List of Tables viii
List of Figures ix
Abbreviations xi
Abstract xii
CHAPTER 1: INTRODUCTION
1.1 Background
of the Study 1
1.2
Problem Statement 5
1.3
Aim and Objectives of the Study 5
1.4
Scope of the Study 5
1.5
Significance of the Study 6
1.6 Organization of the
Research Work 6
CHAPTER 2: LITERATURE REVIEW
2.1
Introduction 7
2.2 Introduction to Database 8
2.2.1 Functions of database 11
2.2.2 Types of databases 11
2.3 Introduction to Database Management System (DBMS) 13
2.3.1 Objectives of DBMS 13
2.3.2 Functions of DBMS 14
2.3.3 Components of a DBMS 16
2.4 Control System Overview 16
2.4.1 Control actions and controller
characteristics 19
2.5 PID Controller 20
2.5.1 Ziegler-Nichols rules for tuning PID
controllers 21
2.5.2.1 Ziegler-Nichols first method 21
2.5.2.2 Second Method 22
2.6 Review of Related Works 23
2.7 Identified Knowledge Gaps 42
CHAPTER 3: MATERIALS AND METHODS
3.1 Materials 44
3.2 Methods 44
3.3 Database Characteristics/ Mathematical Model 46
3.4 Fuzzy Inference
System (FIS) 54
3.5 Database Fuzzy Rule Format 54
3.6 Database Management 56
3.7 Computer Simulations 54
CHAPTER 4: RESULTS AND DISCUSSION
4.1 Results 59
4.2 Comparisons and
Relationships of Functions 61
4.3 Database (A) Management 70
4.4 Database (B)
Management 71
4.5 Comparison of the
Optimized Control and the Conventional Control 75
CHAPTER 5: CONCLUSION AND RECOMMENDATIONS
5.1 Conclusion 79
5.2 Recommendations 80
5.3 Contribution to Knowledge 80
REFERENCES
APPENDICES
LIST OF TABLES
4.1: Comparison
of database structures 67
LIST OF FIGURES
2.1: The feedback control loop 18
2.2: Block diagram of an industrial controller 19
2.3: PID control of a plant 20
2.4: Unit step response of a plant 21
2.5: Transfer function
curve 22
2.6: Closed-loop system with a proportional controller 22
3.1: Method of the work 45
3.2: A scheme for
deriving traffic flow value at each traffic line point. 46
3.3: The block diagram of
the DBMS to be modeled 48
3.4: Design of fuzzy
inference system 51
3.5: Membership functions
input1 52
3.6: Membership functions
input 2. 52
3.7: Membership functions
output 1 53
3.8: Fuzzy plot of
membership functions output 2 53
3.9: Fuzzy rule editor 55
3.10: Conventional model 56
3.11: Optimized model 57
3.12: SIMULINK model showing
subsystem data processing in
conventional model 57
3.13: SIMULINK model showing
subsystem data access rate of the
conventional model 58
4.1: Graph of database
fuzzy controller response. 59
4.2: Graph of SIMULINK
result showing data flow 60
4.3: Graph of fuzzy
controller input rules. 60
4.4: SIMULINK
result comparing data processing with data
access rate in thedatabase 61
4.5: Data tracking 62
4.6:
3D plot of data access Z, database X,
and data flow rate Y. 63
4.7: 3D plot of data process
Z, database X, and data flow rate Y. 64
4.8: Plot showing the
speed at which the database sends and receives data 65
4.9: Plot showing the
speed at which the database saves
and discards data in
the database 65
4.10: Database error
calculation. 66
4.11: Graph of database A
against data size 68
4.12: Graph of database B
against data size 69
4.13: The graph of database
A against data structure 69
4.14: The graph of database
B against data structure 70
4.15 (a): 3D plot of the database A with its data structure. 70
4.16 (b): 3D plot of the database A with its data size. 71
4.17 (a): 3D plot of the database B with its data structure. 72
4.18 (b): 3D plot of the database B with its data size. 72
4.19: 3D plot showing
relationship among database B with data size and
data structure. 73
4.20: 3D plot showing
relationship among database A with data
size and data structure 73
4.21: Graph of fuzzy rule output against Time 74
4.22: Graph of data flow against Flow Time 75
4.23: Graph of data access against Time. 76
4.24: Graph of data flow error against flow time. 77
LIST OF ABBREVIATION
AI ARTIFICIAL INTELLIGENCE
ANN ARTIFICIAL NEURAL NETWORK
ATM ASYNCHRONOUS TRANSFER MODE
CAD COMPUTER AIDED DESIGN
CAI COMPUTER AIDED
INSTRUCTION
CAL COMPUTER AIDED LEARNING
CPU CENTRAL PROCESSING UNIT
DBA DATABASE ADMINISTRATOR
DBMS DATABASE MANAGEMENT
SYSTEM
DRF DYNAMIC RECONFIGURATION ALGORITHM
DTAC DEPLOYABLE TESTBED FOR AUTONOMIC COMPUTING
EFT ELECTRONI FUNDS TRANSFER
FCS FEEDBACK CONTROL REALTIME SCHEDULING
FIS FUZZY INFERENCE SYSTEM
FLS FUZZY LOGIC SYSTEM
GA GENERIC ALGORITHM
IP INTERNET PROTOCOL
MATLAB MATHEMATICS LABORATORY
MYSQL MY STRUCTURE QUERY LANGUAGE
OLTP ONLINE TRANSACTION
PROCESSING
PID PROPORTIONAL INTEGRAL DERIVATIVE
PV PROCESS VARIABLE
QLSM QUERY LOCALITY SET MODEL
QOS QUALITY OF SERVICE
RAID REDUNTANT ARRAY OF INDEPENDENT DISKS
RDBMS RELATIONAL DATABASE MANAGEMENT SYSTEM
RISC REDUCED INSTRUCTION SET COMPUTER
SP SET POINT
STR STRING PROGRAMMING LANGUAGE
TCT TOTAL CYCLE TIME
XML EXTENSIBLE MARKUP LANGUAGE
CHAPTER 1
INTRODUCTION
1.1 BACKGROUND OF THE STUDY
A database-management system (DBMS) is a collection of
linked data and an array of applications for accessing it. A database is a
collection of connected data having an implied meaning that is stored in a
database management system (DBMS). The database is a collection of data that
provides information that is relevant to a business. The basic aim of a
database management system (DBMS) is to proffer a suitable and efficient
mechanism to store and retrieve database information. We use the term
"data" to refer to known facts that can be recorded and have implicit
meaning (Dana, 2006).
A datum, or data unit, is a symbol or combination of
symbols that is utilized to represent anything. The heart of what we understand
by information is the relationship between symbols and the things they
represent. As a result, information is interpreted data — data that has been
given semantics. The practical application of information is referred to as
knowledge. While information may be easily transmitted, saved, and shared, the
same cannot be true for knowledge. Personal experience is required for
knowledge. Referring back to the scientific experiment, a third person reading
the results will have information about it, while the person who conducted the
experiment personally will have knowledge about it (Kurt, 1994).
Database systems are made to handle massive amounts of
data. Data management necessitates both the creation of structures for storing
data and the establishment of methods for manipulating data. Furthermore,
despite system crashes or efforts at illegal access, the database system must
preserve the security of the information stored. If data is to be shared across
multiple users, the system must avoid any unexpected outcomes. Because
information is so important in most organizations, computer scientists have
developed a large body of concepts and techniques for managing data (Zheng,
2000).
Raw
data, which was originally transferred manually from paper to punched cards and
then into data-entry terminals, is now delivered into the system from a variety
of sources, including Asynchronous Transfer Mode (ATMs), Electronic Funds
Transfer, and direct consumer entry through the Internet, under the title Data
Management Systems.
Many businesses rely on complex
database management systems, because it is important to keep them running
efficiently (Kurt et al.,
1994). Tuning in database system
is looking at all aspects of it, including hardware and software, in order to
identify bottlenecks and make query response times as fast as feasible. Noon
and Getta (2016), define tuning as the process of selecting server hardware,
setting up RAID systems, deploying clusters, and customizing operating systems
as well as the database itself for optimal performance. Database administrators
or system administrators are usually in charge of database tuning. Because
database tuning is such a complicated operation, machine learning is being used
to automate it.
Database tuning is the process of
database administrators optimizing a database's performance. In the enterprise, this usually means the maintenance of a
large database management system (DBMS) such as Oracle or MySQL. This includes
optimizing the performance of the database itself as well as the underlying
hardware (Weikum and Chaudhuri, 2000). There has been a surge in
interest in self-managing database and information systems that can
automatically modify their configuration and behavior to meet performance needs
over the last decade (Surajit et al.,
2004).
Although control theory provides a solid
theoretical foundation for the aforementioned issues, applying it to a
self-tuned database management system is far from simple. The inherent
differences between database/information systems and traditional control
systems (mechanical, electrical, and chemical systems) add to the control loop
design challenges. This requires careful mathematical study from a control theoretical
viewpoint (Wang, 1994).
Controlling feedback is crucial when it
comes to database management systems and networks. For example, feedback (also
known as closed loop systems) is used to meet reaction time goals by altering
scheduling priorities, memory allocations, and network bandwidth allocations.
Unfortunately, most computer practitioners construct feedback control on the
fly, resulting in undesirable outcomes such as excessive oscillations or
delayed adaptability to variations in workloads. Control theory is used to
evaluate and construct feedback loops in other mechanical, electrical,
aeronautical, and other engineering areas.
Control
theory is a organized approach to creating closed loop systems that are
constant in the sense that they shun wild fluctuations, perfect in the sense
that they accomplish desired outputs (e.g., response time objectives), and
swiftly settle to steady state values (e.g., to adjust to workload dynamics).
Control theory has recently been applied to the design of various parts of
computing. Control theory has been applied to flow control and the construction
of new Transmission Control Protocol/Internet Protocol (TCP)/IP versions in
data networks, for example (Hellerstein et
al., 2004).
Automatic
tuning has long been an elusive aim in database technology, but it is more
important in current E-services. Studies on self-tuning databases now focus on
long-term performance, which tends to adapt a static perspective of problems.
Due to the inherent variations in workloads and environmental conditions, it is
proposed that maintaining transient performance in such systems is a more
essential issue (Kurt et al., 1994).
The workload performance has also been
modeled and predicted through research. Researchers can achieve adaptivity by
employing a variety of techniques and approaches, such as fuzzy logic, which
employs the Fuzzy Inference System (FIS) to predict buffer-hit-ratio, database
size, and user count. To deal with such dynamics and maintain performance in
self-tuning databases, this work proposes the use of PID controllers based on
feedback control theory. It will go over how to apply relevant
control-theoretical techniques to solve common problems in this field.
The
absence of proper analytical models for complex database systems is among the
key challenges in employing formal methods to analyze such problems. System
identification approaches can help to tackle this problem in part. Self-tuning
database technology, according to one view, should be based on the feedback
control loop paradigm, but it must also be founded on mathematical models and
their right engineering into system components. Furthermore, composing
information services into really self-tuning, higher-level E-services may necessitate
a drastic shift away from complex, highly componentized software architectures
with tight interfaces between RISC-style “autonomic” components. The purpose of this study was to reduce the work load and data
traffic usually observed in database management system by applying a
closed-looped control system.
1.2 PROBLEM STATEMENT
Owing to the difficulty currently experience
in Database Management Systems demanding high volume of human interventions
which often results in poor performance of the system; it is therefore necessary
to develop and introduce a control system using PID controller that allows for
an automated self-tuning Database Management System in real time. Thus, the
volume of human interferences and the resulting associated errors can be
reduced to the barest minimum.
1.3 AIM AND OBJECTIVES OF THE STUDY
The aim of the research work is to apply
control system theory to a Database Management System (DBMS).
The objectives of this thesis are:
·
To review relevant literatures
in the field of study.
·
To model a prototype control
system capable of optimizing the performance of a database management system.
·
To simulate the system modeled
using MATLAB/SIMULINK.
·
To analyze the results gotten
and compare with existing theories to validate the work done.
1.4 SCOPE OF THE STUDY
The study will be limited to the design
and modeling of a control system in a decision-making database system.
1.5 SIGNIFICANCE OF THE STUDY
The research will help students of
electrical/electronic engineering who are carrying out research works on
control of a database management system. And also organizations with database
management systems (DBMS) will benefit in the discarding of manual operation of
their database systems which gives room to irregularities such as fraud,
operational delays, timing errors, etc.
1.6 ORGANIZATION
OF THE RESEARCH WORK
This thesis is
arranged in five chapters:
Chapter 1 presents the introduction and
background information of the study.
Chapter 2 presents the literature review of
the level of Performance improvement in the database management system on the
application of control system. Review of related work, fuzzy logic system and
PID controllers is also presented.
In chapter 3, the methodology and processes
leading to the development of fuzzy inference system are presented. It also
describes the simulation carried out in fuzzy logic toolbox in MATLAB.
Chapter 4 presents the results and discussion
of results obtained from the conventional/analytical and fuzzy logic
method.
Finally, in chapter 5, conclusion and
recommendations for future work is presented.
Login To Comment