Files
Abstract
The need to support large-scale time series data is increasing rapidly. There are emerg- ing Time Series Databases built with conventional relational databases or newer NoSQL databases. The ScalaTion Time Series Database is built on top of its column-oriented in-memory database. ScalaTion is an open-source Scala based big data framework for simulation, optimization and analytics. This database provides support for large-scale stor- age, efficient query processing, pattern matching and a variety of forecasting techniques. Its design goals include the ability to scale up and scale out, and the ability to handle conven- tional multivariate time series. The database provides an easy way to transform a table into a matrix (or vector) which may be used as input for other data science/machine-learning models that are available in ScalaTion. The capabilities are illustrated via a case study of vehicle traffic forecasting. Multiple experiments are conducted to evaluate the performancesof four databases: ScalaTion, MySQL, SQLite, and SparkSQL.