Cost effective failover clustering

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
__________________________________________________________________________________________
Volume: 03 Issue: 03 | Mar-2014, Available @ http://www.ijret.org 218
COST EFFECTIVE FAILOVER CLUSTERING
Nilesh V. Patil1
, Martin P. D’souza2
, Pratik V. Dhiman3
, Parth D. Shah4
1, 2, 3, 4
Student, Computer Engineering, PVG’s COE, Nashik, Maharashtra, India
Abstract
Every company or organization has to use a database for storing information. But what if this database fails? So it behooves the
company to use another database for backup purpose. This is called as failover clustering. For making this clustering manageable
and lucid the corporat people spend more money on buying a licensed copy for both, the core database and the redundant database.
This overhead can be avoided using a database with non-similar platform in which one database is licensed and other may be
freeware. If platforms are similar then the transactions between those databases become simpler. But it won’t be the case with non-
similar platforms. Hence to overcome this problem in both cases cost effective failover clustering is proposed. For designing such
system a common interfacing technique between two non-similar database platforms has to be developed. This will make provision for
using any two database platforms for failover clustering.
Keywords: CEFC (Cost Effective Failover Clustering), DAL (Data Access Layer), DB (Database), HA (High Availability)
-----------------------------------------------------------------------***-----------------------------------------------------------------------
1. INTRODUCTION
In a High Availability Cluster or Failover Clustering
environment the proposed system will allow the user to use
multiple types of Database Management System for whole
cluster. This service will make a provision for organizations
which depend on database cluster to deploy a Database
Management System Independent Failover Cluster.
In a failover cluster, there are two computers (or occasionally
several computers). One (Primary) provides the service in
normal situations. A second (failover) computer is present in
order to run the service when the primary system fails. The
primary system is monitored, with active checks every few
seconds to ensure that the primary system is operating
correctly. The system performing the monitoring may be
either the failover computer or an independent system (called
the cluster controller). In the event of the active system failing,
or failure of components associated with the active system
such as network hardware, the monitoring system will detect
the failure and the failover system will take over operation of
the service. Every company or an organization has to use a
database for storing information. But what if this database
fails? So it behooves the company to use another database for
backup purpose. This is called ad failover clustering. For
making this clustering manageable and lucid the corporate
people spend more money on buying a licensed copy for both,
the core database and the redundant database.
This overhead can be avoided using a database with non-
similar platform in which one database is licensed and the
other may be freeware. If platforms are similar then the
transaction between those databases becomes simpler. But it
won‟t be the case with non-similar platforms. Hence, to
overcome this problem in both cases, cost effective failover
clustering is proposed. For designing such system a common
interfacing technique between two non-similar database
platforms has to be developed. This will make provision for
using any two database management system platforms for the
failover clustering. If the clustering is implemented using this
technique it will provide the same minimum downtime service
to the user at minimum cost.
1.1 Scope
In a HA cluster multiple servers are used to provide a
continuous service to the clients. If some activity internal or
external to the system interrupts the normal working of an
active server then all the clients are switched to the backup
server. If it a database clusters the organization has to maintain
multiple databases in case of failure of active server. These
databases should be of same platform, but it becomes
expensive to buy licensed database. To overcome a difficulty
in HA clusters for providing platform independent DBMS
within same cluster, we will be designing a service which will
be deployed in this cluster. In this way same database cluster
can have multiple types of database management systems.
2. SYSTEM DESIGN
The Cluster system will work in THREE TIER
ARCHITECTURE where a SERVICE acts as a mediator
between the client and the server. Hare the server is in the HA
cluster and Clients are accessing database in cluster. The job
of the service is to monitor the databases and accordingly
directs the clients to the ONLINE database. This service
provides a real time monitoring of the databases. If the service
finds the main database is not active it immediately redirects
all the clients to the redundant backup database. This
switching should take as minimum as possible delay i.e. the

__________________________________________________________________________________________
client show face minimum downtime. This system will only
work if there is something between the client and the server.
2.1 Modules
The CEFC service as a whole is only one process which will
be running in the background. This service will be monitoring
the servers in real-time. The working of the service can be
divided into 3 modules as follows-
2.1.1 Tracing the Queries
First job of the CEFC service is to maintain a real time backup
of the CORE database. As soon as a query is fired from the
application it should be backed up on the redundant database.
For doing this the service will keep a trace of all the queries on
a temporary file which will be stored on the server. This trace
file will contain all queries which modify the OCRE database
and this information will be used to update the redundant
database. In this way a real-time backup is done by tracing the
queries.
2.2.2 Switching the Database
Under the normal operation the system i.e. when the CORE
database is working properly without any issues, the queries
are serviced by the database and the response is given to the
user, alongside the redundant database is backed up. A
failover situation is encountered when by some internal or
external activity the CORE database is shutdown or fails
temporarily. Under such circumstances the users should be
switched to the backup database, this will be done by the
CEFC service with downtime as minimum as possible. This
will be the most critical module of the service as the
Information of the organization is at stake.
2.2.3 Updating CORE Database after Restart
After switching the user to the main database the redundant
database becomes the core for the time instance. But the
CORE database is still resolving from the issues, after this
server is online all the information need to be updated to the
CORE database. This module is handled by the CEFC service
by keeping a log of all the queries that take are fired after the
CORE server went down. The CORE database gets updated as
soon as it comes online and the users switched back to the
CORE server. In this way the normal operation of the whole
system is put back to place.
3. PROPOSED SYSTEM
Once the service is deployed in the system it will start
monitoring the status (online/offline) of the database. On the
current status of the database the service will work in two
distinct ways as mentioned below-
 Under Normal Circumstances: This is when the
CORE database will be active.
 Under Failover: This is when the CORE database
will encounter failure.
o Under Normal Circumstances –
Before failover the main database will be in use by the
client application, at this instance the service will keep
trace of every query in a trace file and update the
respective information in the backup database. In this way
all the information will be replicated in the redundant
database. Consider the following diagram where SQL is
the main database and MYSQL is the backup database.
When SQL server is active all the queries will be directly
communicating with the main database via DAL (Data Access
Layer) the trace of these queries will be kept in the „trace file‟
which will be used to backup these queries on the MYSQL
database. The service by considering that MYSQL is offline
will wait till MYSQL awakes and be online. Once MYSQL is
activated SERVICE will process traced queries on MYSQL.
Fig1 When SQL Server is active
When SQL Server is active all the queries will be directly
communicating with the main database via DAL(data access
layer) the trace of these queries will be kept in the „trace file‟
which will be used to backup these queried on the MYSQL
database. The service by considering that MYSQL is offline
will wait till MYSQL awakes and be online. Once MYSQL is
activated SERVICE will process traced queries on MYSQL.
o Under Failover-
When the main database is disconnected by any means
the job of the service in HA clustering is to switch the
client application to the backup database, in this case it is
MYSQL. Queries are now directed to backup database
but all the information which is processed after the main
database failed till the time it again comes online has to
be updated in the main database. Consider the given
diagram-
MySql
D
A
L
windows
service
SQL
server
Traced
queries
Application
Layer

__________________________________________________________________________________________
Fig 2 When MySql Server is active
Here again the traced files will be used to update the
information once the main database awakes and comes online
again. After updating the queries the client application will
again be switched to the main database.
4. CONCLUSIONS
CEFC Service has enabled us to use multiple types of database
management system(DBMS) in the same cluster. The main
feature of the failover cluster or HA cluster is minimum
downtime is also maintain by the service. If an organization or
company uses expensively licensed database as their core
database and a freeware or cheaper backup database, they will
be saving bulks of cash. This provision is possible only the use
of CEFC Service.
This service also checks the load on a specific server and if
needed the load is balanced by service itself. This is possible
because a cluster consist multiple servers which acts as backup
servers in case of failure of core server. Under normal
operation of the core server if the high load situation is
encountered, the traffic is divided among other redundant
servers in the cluster.
In this age of expensive high end computing our CEFC service
allows even a small organization to deploy Cost Effective and
Efficient Failover Cluster.
REFERENCES
[1] Greg Pfister: “In Search of Clusters, Prentice Hall”
[2] Evan Marcus, Hal Stern: “Blueprints for high
Availability: Designing Resilient Distributed Systems,
John Wiley & Sons”
[3] Chee-Wei Ang, Chen-Khong Tham: “Analysis and
optimization of service availabitity in a HA cluster with
load-dependent machine availability”
Sql
server
D
A
L
windows
service
My
Sql
Traced
queries
Application
Layer

Cost effective failover clustering

More Related Content

What's hot (6)

Viewers also liked (20)

Similar to Cost effective failover clustering (20)

More from eSAT Publishing House (20)

Recently uploaded (20)

Cost effective failover clustering