Administration Guides

This section describes the procedures to setup a single-node or multi-node Ophidia cluster, as well as the steps to install and configure the single Ophidia software modules and the commands to manage the platform. If you are just interested in getting Ophidia source code, RPMs or Virtual Machine Images visit the download section on Ophidia website.

Before starting with the actual setup and deployment, it is important to briefly understand the key architectural blocks composing an Ophidia instance.

In the most general deployment, an Ophidia cluster can be composed of different types of nodes (Linux OS is required on each):

  • a client node
  • a single server node
  • one or more framework nodes
  • one or more analytics nodes

This node hierarchy has been designed to be very flexible in order to adapt the deployment based on the cluster environment. In fact, if users have direct access to the machines of the environment, client and server components can share the same physical node. Moreover, framework and analytics components could be also deployed on the same node. In this configuration the node is called a Super node. For training purposes the component can also be deployed all on the same node (called single node deployment).

If you are intereseted in the setup of a simple single node deployment refer to single node setup. This configuration is only suggested for testing and training puroposes since its scalability is limited by the node resources. For a full multi-node production deployment able to scale over multiple resources refer to cluster setup. This second configuration is the one recommended for HPC and Cloud environments composed of multiple nodes.

Finally, if you are interested on the installation, configuration and management of specific Ophidia components, you can refer to the following links.