Hortonworks architects vertically integrate the projects within our Hadoop distribution with YARN and HDFS in order to enable HDP to span workloads from batch, interactive, and real time—across both open source and other data access technologies. In HDP 2.2, we deliver work to vertically integrate Apache Storm, Apache Accumulo and Apache HBase so that all of those long-running services run in Hadoop on YARN via Apache Slider.
The Apache Slider community recently released Apache Slider 0.60.0. Although there are many new features in Apache Slider 0.60.0, four new innovations are particularly notable:
Here are some details about those four important features. For the complete list of features, improvements, and bug fixes, see the release notes.
Existing long-running applications can be difficult to update for deployment on a Hadoop cluster. The new framework in Apache Slider provides a way to leverage YARN’s resource management capabilities. Now cluster administrators can onboard existing long-running data access applications, without changes to their code or API implementations. This allows consolidation of all edge clusters onto a single, unified Hadoop cluster.
Scale applications in Slider on demand with the ability to increase and decrease component instance containers. Now you can easily suspend or resume applications with Slider. These changes provide new application elasticity and agility, which improves cluster utilization and reduces total cost of ownership.
Apache Ambari now has a View called “Slider App View.” This shows all the apps running through Slider, with their related details, providing a single, integrated management interface for all Slider-enabled applications.
The Slider community plans to develop some exciting new capabilities for the next release, including:
We want to thank the Apache HBase, Accumulo, and Storm communities for their work to enable those powerful long-running applications with Slider. We also thank the Apache Ambari community for providing the management view for Slider.