Using Timely as a time series database

Now that we are able to transform raw information into a clean series of Twitter sentiment with parameters such as hashtags, emojis, or US states, such a time series should be stored reliably and made available for fast query lookups.

In the Hadoop ecosystem, OpenTSDB (http://opentsdb.net/) is the default database for storing millions of chronological data points. However, instead of using the obvious candidate, we will introduce one you may not have come across before, called Timely (https://nationalsecurityagency.github.io/timely/). Timely is a recently open sourced project started by the National Security Agency (NSA), as a clone of OpenTSDB, which uses Accumulo instead of HBase for its underlying storage. As you may recall, Accumulo supports cell-level security, and we will see this later on.

Storing data

Each record is composed of a metric name (for example, hashtag), timestamp, metric value (for example, sentiment), an associated set of tags (for...