Field Notes: Apache Ambari Meetup at Hortonworks
On April 2nd, Hortonworks was excited to host the very first Apache Ambari Meetup. Thanks to all those who came along in person and virtually for a lot of vibrant discussion. If you would like to get involved in future Ambari Meetups, please visit this link. We are well on the way to making Hadoop management ‘dead simple’.
We have embedded the sessions below with some notes:
Overview and Demo of Ambari, Yusaku Sako, Hortonworks
- This session covered Apache Ambari’s mission to “Make Hadoop management dead simple”, Ambari’s 4 major roles: 1) Provision, 2) Manage, 3) Monitor, and 4) Integrate, emphasized that everything that Ambari’s Web Client does is done thru Ambari’s REST API (100% REST), presented high-level architecture, and a live demo on how to provision, manage, and monitor a Hadoop cluster using the latest Ambari 1.2.2 release.
- The project website can be found at http://incubator.apache.org/ambari, with all the info about Ambari within. There was encouragement for everyone to contribute to Ambari’s success through filing bugs, participating in mailing list discussions, providing feedback and direction, writing documentation, submitting patches, etc.
- APIs and SPIs of Ambari (How to Integrate with Ambari). Tom Beerbower (Hortonworks)
- Tom presented on the details of Ambari’s REST API and a live demo of the API working in action.
- The SPI (Service Provider Interface) and how its plug ability allows various integration scenarios were explained.
- This was a great lead up to the next presentation by Teradata, who integrated Hadoop monitoring to their management software Teradata ViewPoint using Ambari’s REST API and SPI.
- Teradata ViewPoint Hadoop Integration with Ambari. Steve Ratay (Teradata)
- Steve presented on how Ambari is a key enabler for integrating Hadoop monitoring to ViewPoint. Without Ambari, integration of Hadoop monitoring to ViewPoint would have been difficult (need to collect metrics from a number of sources spread across different technologies and formats, such as Ganglia, Nagios, JMX, and screen scraping Hadoop’s native web UI). Ambari REST API provides a central place with a single, consistent data format (JSON) with powerful querying capabilities. As Steve put it, “Ambari to the Rescue!”.
- The highlight of the presentation was a live demo of Teradata ViewPoint with a plethora of Hadoop metrics exposed through Ambari REST API behind the scenes. Steve said that the integration only took a couple of months of effort by a couple of people.
- Ambari Futures. Jeff Sposetti (Hortonworks)
- Jeff presented what’s in store for Ambari in the future starting with what’s currently being worked on for the 1.3.0 release, as well as beyond…
- There was a lot of interest in the room around extensible stack definitions to integrate any Hadoop ecosystem component.
- Also a concept of Cluster Blueprints was shared, where it would allow “zero-touch” and headless installs of Hadoop clusters.
You can watch all 2 hours of proceedings at this link. Once again, thanks to everyone who attended and took part in a great conversation – see you next time!