IMPROVING DATABASE AVAILABILITY THROUGH FAULT MONITORING AND EARLY DETECTION

IMPROVING DATABASE AVAILABILITY THROUGH FAULT MONITORING AND EARLY DETECTION

ABSTRACT

Data storage and retrieval are at the centre of today’s applications. Currently, no organization can effectively carry on its activities without one form of database or the other. Indeed, one of the key challenges of information technology organizations is how to ensure quality and reliable database administration. Consequently, if the database and the server that support the applications are unavailable it could lead to considerable costs in terms of efficiency, loss of  revenue as well as damage to reputations and goodwill. The aim of this study is to compare three open-source database monitoring system tools that are cost effective and  able to notify the Database Administrator of any fault in the system. They are Nagios, Zabbix, and Cacti. For the purpose of this study, an empirical client-server architecture environment was created with Red Hat Enterprise Linux 6.5 Operating system in a Local Arrea Network using a switch to which a MYSQL database server, monitoring tool server and host computers were connected.  The results showed that the three monitoring tools studied were able to monitor database availability and could send prompt notification to the Database Administrator via email, with varying performances. Nagios was shown to be superior to the other tools as it was able to detect and send notifications about database failure faster than others. While Nagio takes about 75 milliseconds to report Network failures, Cacti and Zabbix take about 102 and 145 milliseconds, in that order to report Network failures. In the same vein, Nagios reported Database faults in 35 milliseconds while Cacti and Zabbix reported the same fault in 123 and 157 milliseconds respectively.  Nagios was also shown  to be able to handle 22 different services concurrently while Cacti and Zabbix could only handle 15 and 12 services at the same time.