Scalable and user friendly user interface for time-series analytics for OpenTSDB

Abstract

OpenTSDB is a fast and reliable database used worldwide. While it has numerous advantages, its current web user interface is simplistic and not interactive, wasting the time that takes to perform a specific task . This thesis focuses on the implementation of a more reliable and interactive architecture using a Model, View, Controller architecture while considering visual analytics, NodeJS, websockets, Python, and R. The nodejs server is proposed as a solution. It has four different built-in connectors that obtain and transform data from OpenTSDB. We will show a connector from OpenTSDB to the NodeJS server (nodetsdb), OpenTSDB directly with the client javascript (nodetsdb-client), OpenTSDB to Python, and OpenTSDB with R. After implementing and testing all the connectors we discovered that the connector from OpenTSDB to NodeJS is the fastest one, retrieving one month of data points in less than sixty ms.