Hadoop's supporting cast
Hortonworks put faith in open source management tool Apache Ambari
Hadoop vendor Hortonworks have unveiled their newest
version of their open source distribution, Data
Platform 1.2, welcoming management tool
Ambari into the stack.
There’s plenty of established projects (like Hive, Pig and HBase) that sit aside the core Hadoop technology, adding in new functionalities. But a frequent enterprise criticism of Hadoop is its difficulty to manage a large number of clusters, especially without being able to set up a graphical interface easily.
Apache Ambari appears to be an alternative to this problem, with the Hortonworks team describing it as “a key operational component within the Hortonworks Data Platform (HDP), which helps make Hadoop deployments for our customers and partners easier and more manageable.”
Hortonworks are placing emphasis on its stack being “the industry’s only complete 100-percent open source platform”, and Ambari slots into that. With competitors MapR and Cloudera initially only offering closed source/fee-based management tools, Hortonworks could gain enterprise trust for the malleability of their open source option.
The Java-coded, web-based tool allows inexperienced sysadmins to provision Hadoop clusters through a step-by-step wizard. The UI makes it possible to track MapReduce jobs and monitor performance metrics across the cluster, making it quicker to locate bottlenecks. Metric collection is controlled through Ganglia, while Nagios is responsible for sending messages to alert the user of any problems.
Ambari’s RESTful API means users can initiate commands and also integrate with third-party system management tools in the datacenter.
By Apache standards, Ambari has been progressing quickly in the incubator, “shipping often and early releases” according to the Hortonworks team, who predominantly work on the project. Several Apache Hadoop stablemates are supported within Ambari, including Hive, HCatalog, Pig and Zookeeper.
Ambari has been available in Hortonworks Data Platform since June’s 1.0 release, but in tech preview only. More details about the release can be found here. Despite Hortonworks' decision to stick with the mature Hadoop 1.x codebase as the basis was thought to be conservative by some, but their willingness to innovate isn't in question.