The usage of Internet has increased rapidly over the past few years and is expected to increase in the coming years as well. This increase in usage also increases the possibilities of malicious endpoints. Middleboxes have become a basic component of the networks. The network traffic has to be able to pass through these middlebox(es) before reaching the target endpoint. These middleboxes can behave very differently depending on their implementation logic and hardware resources. It is important to know the current behaviour of middleboxes deployed in the Internet to be able to pass the traffic through middleboxes successfully. This becomes even more crucial when new services are deployed over the Internet. The aim of this thesis is to create a mechanism to explore the behaviour of middleboxes in the current Internet deployments. The thesis describes a middlebox test system, named "Obstacle Tester", that consists of a client and a server where the client usually resides behind a middlebox and the server is connected to the Internet directly. The client performs a number of tests against the server to detect the collective middlebox behaviour in the path toward different traffic flows and characterize the behaviour. The test results are saved in the database at the server which allows post processing of data and discover the behaviour of deployed middleboxes over a larger network footprint.