Sei sulla pagina 1di 7

Application of MySQL Cluster to Build Two-

Way Distribution of Distributed Database


with Application of Fail Over and Load
Balancing Method

Pascal Jezequel Mario Campone Wilfried Gouraud


Catherine Guerin-Charbonnel Christophe Leux
Gabriel Ricolleau Loc Campion

Abstract - An organization or server with the same database.


company must have in it an To accomplish this, we use the
adequate information system. Fail Over methodology with
Information systems are usually Replicate Two-Way Database and
used to facilitate employees / Load Balancing with the
clients / stakeholders who play a assumption that the workload of
role in it to seek information. each server can be shared. With
With the existence of complex the application of this method
information systems required a obtained the results of the
database storage to support the server one and the other can
running of the information backup each other data, and also
system. In reality, often when the server one down it can
encountered a condition where be replaced by other servers. So
the information system when it can be concluded that the
accessed by the client implementation of Database
sometimes failure. Failure is like Replication with Failover and
a failure on the required system Load Balancing can provide
against the database. So the solutions especially for
failure of the database, the companies or organizations that
automatic is very closely related are in need of availability of
to the server so it appears like server data, in this case that is
the failure. Usually it happens High Availability.
because the server is off or down
or the network is disconnected Keywords-Replication,
so that the client can not access MySQL Cluster, Fail Over, Load
the existing data on the Balancing.
database server. Therefore the
solution of the problem that is I. INTRODUCTION
required at least two servers
(Master-Slave) for database Problems here the author took
storage is High Availability. So from the results of previous journal
that the client in accessing the reviews, such as the journal entitled
server database if the failure "Population Data Distribution In
encountered will be directly Gorontalo City Using Database
replaced or served by another
Replication Methods With Single The three main nodes that are
Master Replicated Techniques" components of MySQL cluster
Where from the above journals include:
there is a problem that is the A pair of Data Nodes that are
availability of data is still centralized, used to store all transaction data on
then completed by the author of the the MySQL cluster and the data is
journal with the method that the replicated on this node, the
authors sense is still lacking. So this Management Server Node is used to
writer thinks there is a need for a control the processes on the system
development technique that can when it is turned on. In addition, this
solve the problem with several node is also used to identify any
methods that are closely related to configuration changes, MySQL Server
the distributed data base. API Node is an interface used by the
Based on the issues discussed client to access data and manipulate
in this journal, can be drawn some data on the cluster system.
discussion of issues that refers in the
theme or title (Building Replication 2.3. Failover
Two-Way Distributed Database with Failover is the process of
Application of Fail Over and Load automatically switching a server
Balancing Method) as follows, when another server is experiencing
namely: Understanding the concept a problem or malfunction or can also
of two-way Replication in database be said to be duplication and data
design Distributed, two-way backup.
replication configuration with This process can occur without
combination of Failover and Load stopping the running process so that
Balancing methods, the advantages the movement is done by the cluster
of two-way Database Replication with system itself. This type of clustering
Failover and Load Balancing is more intended for data duplication
methods, Implemantation of Two-way so that the client can still access the
Replication with Failover and Load data information service even though
Balancing methods. the server side is actually being
interrupted.
II. METHODOLOGY
2.4. Load Balancing
2.1. Two-way Replication Load balancing is a technology
Database replication aims to of sharing internet traffic or work
duplicate a server, so that the processes on multiple servers using
database has duplicates on other hardware and software in a computer
servers. network and can be run as a router.
Two-way replication means not This clustering system serves to
only one server can duplicate data to distribute the traffic load on two or
another server, but one server and more connection lines in a balanced
the other can duplicate each other's way so that traffic can run optimally
data and in this context the roles of with a small response time and
both servers are the same. optimal process.
Load balancer will receive a
2.2. MySQL Cluster traffic from the client and then the
traffic is continued to some other Figure 1. MySQL Cluster Topology and
server. The traffic will be sorted by Load Balancing
client request and server condition.
When only one client requests, the In Figure 1 it can be
process happens only on one server. interpreted that Managent Node
Whereas when exceeded, the load Sever acts as a controller or
balancer will divide the process in a controller than other server nodes.
balanced manner for each server.
2.6. Physical Data Model (PDM)
2.5. Network Architecture This is a database scheme
Here will be described the which will be replicated later.
architecture or network scheme In Figure 2 is a Physical Data
MySQL Cluster and Load Balancing: Model scheme, here the authors use
a subdistrict database consisting of
several tables including: Accounts,
Profiles, Employees, Positions,
Salaries, Allowances, and others.
III. RESULTS AND DISCUSSION Figure 4. Initialization process on
Here the writer first discuss the Node Server
connection problem between node-
node sever to be able to duplicate In Figure 4. It has been seen that
data between server one and other both server nodes have been
server, as below: initialized in order to be recognized
by Management Server.
3.1. Connection between Node Sever
and Management Server 3.2. Two-way Replication
The authors apply a two-way
relication where the two servers can
duplicate data, either server one or
the other server.

Figure 3. View Connection between


server nodes in Management Node

If it looks like picture 3 then it is


certain to do two way replication.

Next do initialization on the server


node to be recognized on the
management node, as below:

Figure 5. Replication Results

The data inputted by Node One


automatically replicates on the
Second Server Node. Terms to be
able to replicate here is the second
database server must be NdbCluster
engine because this replication using
MYSQL Cluster so if you use the
default built-in MySQL engine
MyIsam then the replication process
will not be able to run. Figure 5. As
proof that the query results
successfully entered.

3.3. Failover and Load Balancing


Figure 8. Conditions when no
Here are the results of failover incoming queries
testing:
For this we can look at Session Max,
where the value is 0, because there
is no query yet.

Figure 9. Conditions after the data


Figure 6. The second condition of the query occurred
Node Server is on
When the data query occurs, it can
Node one on Load Balancing shows be seen in Session Max will change
that DownTime looks 12m53s when the value to 1 where this value 1
it is on live condition. indicates the occurrence of the data
query process.

IV. CONCLUSION

Based on the results of


research, it can be concluded that to
run the replication of both server
nodes one and two must have been
Figure 7. Conditions when Node one running first and can be connected
is turned of with Management Node. After the
Management Node Getting the
When Server Node One is turned of position of the two Nodes, then it can
then there is a change on DownTime be run two-way replication.
to 17s. At the condition of Node one From the test results from the
is dead then the data query result server Node One obtained value at
data when executed by the user will the time of the data query process
temporarily be moved to the server then the query results are captured
that is still running, in the picture by the second Node database and
above then, the query data will be vice versa. So Load Balancing here is
moved to the Node two server to be very helpful server performance
stored and the replication process because it can play a role to share
will run if the Node server one live the burden between the two servers.
back. Suggestion of development proposed
by the writer that is for the next can
be done application of Failover Database and Settlement Strategy
method with HeartBeat. for Synchronizing Data". Journal of
IPTE-KOM, UGM, Vol. 14, No. 2, 2012.
BIBLIOGRAPHY [4] Deshanta Ibnugraha, Prajna.
"Design and Development of Failover
[1] Eddy Purwanto. 2012. System In MySQL Using Heartbeat
"Comparison of Replication Strategy and MySQL Native Replication to
in Distributed Database System". Support Online Data Availability".
Journal of Bina Darma University Journal of Information Technology of
Palembang, Vol. 5, No. 10, 2012. Telkom Polytechnic, Vol. 1, No. 3, May
[2] Abdul Gani, Taufiq. "MySQL 2012.
Cluster Performance Analysis Using [5] Silitonga, Parasian. "Database
Load Balancing Method". Journal of Replication In The Academic Data
Electrical Engineering, Vol.11, No. 4, Processing System of St. Thomas
August 2015. Catholic University". Journal of TIME,
[3] Raentantyo Wardoyo. 2012. Vol. 3, No. 1, 2014.
"Code Diferences in Rancanagn

Potrebbero piacerti anche