This blog post was authored by Mitchell Tracy.
In Vertica 8.1.1, we introduce support for Cloudera Manager. Cloudera Manager is a platform that Hadoop administrators can use to manage their Hadoop cluster. It allows them to see the hosts associated with their cluster, and the different Hadoop services running on the cluster. Cloudera Manager also allows you to monitor hosts to see the amount of memory consumption, disk I/O, and other metrics. It is an invaluable tool for Hadoop administrators managing a CDH Hadoop Distribution.
The following graphic illustrates the overall view of a cluster:
However, many Hadoop administrators using Vertica SQL on Hadoop don’t have visibility of Vertica clusters; they can’t see which machines comprise the Vertica clusters, nor can they see the relative resource consumption of Vertica compared to Hadoop services running on the same machine.
To address this issue, we developed a Vertica Custom Service Descriptor (CSD) that allows Hadoop administrators to install a Vertica Service in Cloudera Manager. During the Vertica Service installation, users specify the machines that are running as part of the Vertica database. After the installation, users will see Vertica listed as a service running on the specified machines, and can restart the database as a whole, as well as restart the database on particular machines.
The following graphic shows the different options you can take on a Vertica cluster:
Additionally, this release lets you use the Vertica CSD with the Cloudera Static Resource Allocation Wizard. This allows admins to specify a breakdown of how much memory should be assigned to Vertica versus other Hadoop services.
The Vertica CSD is packaged as a jar file in the Vertica RPM, and more specific installation and usage directions are in the Vertica documentation
. The CSD will allow users greater visibility and usability of Vertica, particularly in the world of high performance SQL on Hadoop.
For more information, see Integrating With Cloudera Manager
in the Vertica documentation.