Network analytics has become increasingly popular in recent years. Various graph systems have been developed for analysing networks, while network data is still largely stored and managed in relational database systems in the first place. As two separate systems are often used to manage and analyse network data, it not only increases the difficulty for users to learn and maintain these different systems simultaneously, but also impedes performing more sophisticated analysis on relational and topological properties of network data. Aiming to tackle these issues, we present Rogas in this paper, which is a declarative framework that allows the user to formulate analysis queries naturally without thinking about the tedious implementation details of graph algorithms and query processing.