Welcoming any data, in any format, in any version, at any time...
By Henrik Bærbak Christensen, Department of Computer Science, University of Aarhus.
Karibu is a high quality data collection framework aiming at data integrity, flexibility, and scalability. It is developed for environmental data collection from smartphone apps as well as stationary sensors, and presently used as the backbone for several research projects on-going at Computer Science, University of Aarhus.
Karibu consists of a client library that can be deployed on a Java enabled device (Android phones for instance), and a server side daemon responsible for storing incoming data. In addition, a Karibu system must have a messaging system (RabbitMQ) and a database (MongoDB) deployed.
To get started on the tutorials, you will need Java 1.7+, Ant 1.8+, Ivy 2.4+, and to simulate the distributed environment using the provided virtual machines you will need a way to run VMWare virtual machines, such as VMWare Player for Windows and Linux, or VMWare Fusion for OS X.
Also, Ivy will need to download modules so you will have to be on-line to make the tutorials work.
Using Karibu for your particular data collection needs requires two important steps
-
Configuring Karibu to handle your particular type(s) of data. This involves writing implementations of the
SerializerandDeserializerinterfaces and testing that they work correctly together with Karibu. Tutorial: Hello World. -
Setting up the distributed Karibu environment which includes RabbitMQ, MongoDB, Karibu daemon(s), and the clients. Tutorial: Quick Start.
As Karibu is designed with a lot of flexibility, these two tasks can actually be made in any order.
The Hello World tutorial focus on the architecture and the code and configuration options of Karibu in order to develop the binding to your particular types of data.
The Quick Start tutorial focuses on setting up a fully distributed environment that contains the core aspects of a full production systems but demonstrates it on a provided, fixed, data type.