Installation Guide

Big data Cumulus-Linux installation guide

Issue link:

Contents of this Issue


Page 4 of 26

B I G D AT A W I T H C U MU L U S L I NU X 5 Figure 2. Hortonworks Data Platform and Cumulus Linux The Power of Open Networking The path to building a modern Layer 3 network is made easier and more affordable by Cumulus Linux. Rather than locking yourself into a proprietary solution from a specific vendor, you now have choice through open networking. Cumulus Linux gives you the choice of network hardware from a large HCL of vendors at an attractive and affordable price, allowing you to build a Layer 3 network and grow as you need. Cumulus Linux allows you to save capital expense while building larger, higher-performance leaf and spine network fabrics. Since Hadoop is moving from being primarily batch focused to incorporating real-time and streaming data processing, a non-blocking network fabric powered by Cumulus Linux that can provide comprehensive statistics and programmability is beneficial. Running Apache Hadoop with Cumulus Linux The ideal network design for running Hadoop takes a Layer 3 routed model that can be scaled out by adding spines and leafs, or uplinks. The more spines and uplinks that are utilized provide more redundancy, more cross-sectional bandwidth and make the failure domains smaller. A Layer 3 Clos fabric can be designed to be non-blocking or oversubscribed. From an East-West traffic standpoint, full bandwidth is available between any pair of servers. You can achieve faster recovery with plenty of core bandwidth to re-replicate after failure. The leaf and spine network architecture also provides a consistent distance across the cluster for a low and deterministic latency where every server is equidistant within a pod and equidistant to other pods.

Articles in this issue

Links on this page

view archives of Installation Guide - Big data Cumulus-Linux installation guide