Monitor Hadoop, hbase with ganglia

Ganglia is a monitoring server, cluster of open source software, can be used to graph the last one hours, the most recent day, last week, the last January, the last year of the server or cluster CPU load, memory, network, hard disk and other indicators.

The power of ganglia is that the ganglia server isable to collect data from all clients on the same network segment through a client, ganglia the cluster server can collect all client data from its subordinates through a single server. This system is designed to indicate that a single server can manage tens of thousands of machines through different tiers. This function is unmatched by other mrtg,nagios,cacti.

Ganglia is an open source real-time monitoring project launched by UC Berkeley to measure thousands of nodes, providing system-static data and important performance metrics for cloud computing systems.

The ganglia system basically contains the following three parts.

Gmond:gmond runs on each computer, and it primarily monitors the collection and delivery of metrics data (such as processor speed, memory usage, and so on) on each machine.

Gmetad:gmetad runs on a host in cluster as a Web server, or for communicating with Web server.

Ganglia Web Front end: The Web front end is used to display the Ganglia metrics chart.

Hadoop and HBase itself support the ganglia very well. With a simple configuration, we can display some of the key parameters of Hadoop and hbase in the form of a chart on the ganglia Web console. These are a great help for us to gain insight into the internal system state of Hadoop and HBase.

Monitor Hadoop, hbase with ganglia