Last weeks, we’ve started up the database management system that will hold on whole information of the SmartPolitech project, setting up Apache Cassandra in a single cluster organized, for the moment, in 3 nodes.
Cassandra was selected because its capabilities to store real time data from several sensors that currently are part of the infrastructure of SmartPolitech, as well the fast retrieval of information from large data sets, horizontal scalability (when we need it) and the integration of Hadoop as support for MapReduce programming model.
We had been running different tests to check the performance offered by this type of NoSQL databases.
To check the data distribution, we’ve used a Raspberry Pi, which was outside the university intranet, and was running different Python processes for periodic writes on a node; immediately, we made queries in a different node to confirm that the information has been propagated by the cluster. Later, we brought down a node and restarted it to check the data load balancing.
Next weeks, we have pending different task ranging from connect all sensors to a Cassandra stress test.