Abstract. Brazil has a significant amount on a wide range of data about weather and climate, collected from sensors or calculated by numerical models, and are very important for historical reasons for the understanding of climate change and prediction of extreme weather events. This data represent different physical measures, have different temporal and spatial scales and is stored in different formats; there is no unified way to discover which data is available and under which conditions it can be used. In this paper we describe the architecture of the Brazilian Weather and Climate Virtual Observatory, a set of software tools that works as a Virtual Observatory (VO) to allow weather and data metadata discovery and data access and processing. The VO will be partially deployed in a private cloud; reasons and benefits of doing so will also be explained.