Page 131 - ASBIRES-2017_Preceedings

INCREASING THE SPEED OF SEARCHING PROCESS IN MYSQL DATABASE FOR DYNAMIC DATA
modeling and query language (SQL).
(Relational Database Model | Wikipedia)
When the amount of data stored on tables
are increasing rapidly, in order to store large
amounts of data NoSQL came into
existence.
NoSQL is a non-relational database
management system, designed for
distributed data stores that require very large
data storage on a large scale. Database
framework and computerized NoSQL are
used mainly because of their speed,
scalability and flexibility. These types of
data storing may not require fixed schema,
avoids join operations and typically scale
horizontally. NoSQL databases provide a
mechanism to store and retrieve data using a
model that is more than the tabular relations
used in relational databases like MySQL. (1
what is NOSQL | Yoda Learning) There are
many database engines following the
NoSQL technology and having non-
relational models. Some of them are
MongoDB, HBase, Cassandra and
ElasticSearch. (What is NoSQL? Why
NoSQL? When NoSQL? | Bodhtree Blog)
Elastic search is a real time search engine
which uses a NoSQL database and also a
freeware. (Sense Documentation | Elastic)
(Elastic search: RESTful, Distributed Search
& Analytics | Elastic)
A data cube is a three or higher
dimensional array of values which are
commonly used to store a time series of
data.
3 METHODOLOGY
NoSQL database can be used where a
bulk amount of data can be stored. NoSQL
allows high speed searching and higher
capacity of data which answers to the
available problem. Considering the features
and capabilities of the selected database
engines, Elastic search was selected.
Considering the efficiency, the system
can be successfully migrated into Elastic
search. But even the issues of reliability,
capacity and speed will be resolved by this
solution; still needed the same amount of
121
physical storage space as before. This
requirement does not meet by the available
infrastructure currently. It is suggested to
use data mining techniques to filter and
cluster relevant data. Here data mining can
be used to filter important statistical data
that are needed in calculations and just
remove the rest. Then the mined data will be
stored as data cubes where the data can be
easily retrieved and stored using less space.
The data and their fetching patterns are
analyzed and found that multiple data cubes
can be created. The attributes needed for
that, as well as sample reports that the
business can be given after implementation
was studied. Elastic search and data cubing
can be applied without any restrictions from
TravelBox’s end at least as a structure
without going into the actual
implementation.
Even though the performance issue of
search functions when data fetched from the
MySQL tables could be solved with the
selected solution still there exist the storage
issue. Since MySQL didn’t have the
capability of handling the bulk amount of
data the tables were truncated time to time.
Therefore, the lack of storage space had not
experienced as a critical issue. But with the
use of NoSQL database and Elastic search,
the system will be capable of handling the
actual amount of data without frequent
trancations. Therefore, when the data start to
be loaded into the database a storage issue
will be emerged where the disk space
becomes not sufficient for the bulk amount
of data which will be approximately 1431
GB (Giga Bites) per year without
considering increases in transactions in peak
seasons. This could be twice as the above
mentioned value when the peak seasons are
considered. Therefore, either the storage
capacity should be increased enormously or
should utilize the existing storage to handle
the bulk data efficiently. As a solution it is
suggested to use data cubes to store data in
the NoSQL database. Since data cubes store
values in a multi-dimensional structure that
would be possible to keep the statistical data
in a fewer number of indexes by compacting