Sentiment Analysis techniques have been largely applied to Tweets, newsgroups and Social Networks in general, with several applications in sociological studies. Users tend to comment and express their opinions much more genuinely on Social Networks, as if their natural filters were somehow lifted. In particular, complaints regarding malfunctions of specific services are often filed in form of public comments or Tweets, on the official accounts of the Service providers. In some cases, people just express dissatisfaction regarding services on their own accounts, and use hashtags to better identify the specific topic they are referring to. In this paper, a framework for the analysis of Tweets is proposed, with the specific objective to identify malfunctioning of essential services, such as water, electrical, gas or public illumination. Since the number of comments and Tweets to analyse is considerable, a microservices based architecture, with Docker containers and Kafka queues, has been created. This allows to define a scalable and parallelizable architecture, whose characteristics can be adapted to the number of Tweets to be analysed, which are in turn treated as a continuous data streaming.