On December 18th, 2014, Hortonworks presented the last of 8 Discover HDP 2.2 webinars: Apache HBase with YARN & Slider for Fast NoSQL Access. Justin Sears, Jeff Sposetti and Mahadev Konar hosted the last webinar in the series.

After Justin Sears set the stage for the webinar by explaining the drivers behind Modern Data Architecture (MDA), Jeff Sposetti and Mahadev Konar introduced Apache Ambari and discussed Ambari innovations now included in HDP 2.2:

  • Configuration Enhancements, including Versioning & History
  • Ambari Administration, including Views Framework
  • Ambari Stacks "Stacks Advisor"

Here is the complete recording of the Webinar

Here are the presentation slides.

We're grateful to the 204 attendees who joined the HDP 2.2 webinar and asked excellent questions. This is the list of questions with their corresponding answers:

Question Answer
What is the upgrade path from HDP 2.0 to HDP 2.2? Are there instructions on how to upgrade Ambari 1.6.1 to 1.7? This document includes information on upgrading Ambari and upgrading the HDP Stack.
How does Ambari fit in an enterprise that is embarking on AWS platform? Ambari can deploy Hadoop onto hosts that are physical, virtual or cloud, such as AWS. The infrastructure is really transparent to Ambari, as it only has to deal with the guest operating system, regardless of infrastructure.
Does Ambari configuration management only apply to changes made through Ambari? If changes are made at the command line, will they be captured in Ambari? Ambari manages a set of configuration types for the Services in your cluster. If you make local changes to those files on individual hosts, Ambari will overwrite with the configurations Ambari has stored to maintain consistency across the cluster.
When will HBase and Storm on Slider be integrated into Ambari? A deeper Ambari-Slider integration is in the works but for today, those applications on YARN can be deployed via the Slider view that is included with Ambari 1.7.0.
Can we see how those recommended values are calculated? The Stack Advisor code (which encapsulates the logic) is available in the Stack definition (hosted on the Ambari Server) and is readily readable.
Does the 'stack advisor' perform its calculations based on hardware, cluster capacity or other industry best practices? Depends on the configuration. The calculations can take into consideration host memory, other co-located components and other configuration values to make its recommendations. The logic for the recommendations in general is based on input from the Hortonworks' team of experts.
Are there any Out of Box Views which HW ships to customers? Ambari 1.7.0 comes with two Out-of-the-Box views: Jobs (for viewing Hive Queries executing on the Tez engine) and Slider (for deploying Slider-based applications from Ambari). There are more views contributed in the community. https://github.com/apache/ambari/tree/trunk/contrib/views
Is it possible to add more nodes using Ambari? Yes, Ambari exposes a way to "Add Hosts" to a cluster. During the process, you can also designate which components (such as DataNode or NodeManager) to install on the new hosts.
Can you explain how Ambari helps during initial installation? Ambari exposes two methods to perform a cluster install: via a wizard-driven experience, and an API-driven method called Ambari Blueprints. In either case, Ambari helps you choose the services to include in your cluster, layout the components, validates the layout (to confirm you have the right number and correct co-locations of components) and recommends plus validates configurations for those items. Once you have that in place, Ambari will install the packages, push the configurations and start the components. At the end, Ambari will run a smoke test against the services to confirm they are operating OK.
Is Ambari mainly for the Administrator of the HDP? Ambari is used today by Hadoop operators to provision, manage and monitor clusters. Depending on the set of views you use and the Ambari deployment choice, you will likely see that expand to cover users beyond Hadoop operators.
Can I select specific services from the wizard during installation? When you go through the Install you may select specific services (such a HDFS, YARN, Storm, etc) you wish to install.
Does Ambari 1.7.0 support Apache Spark? Not at this moment. Hortonworks has made a Spark Technical Preview available here: http://hortonworks.com/hadoop/spark/
distributed by