Many multilevel security relational models have been proposed and different models offer different advantages. In this paper, we adapt and refine some of the best ideas from these models and add new ones of own to extend our Multilevel Security with Key-polyinstantiation (MLSK) relational model. MLSK now supports relational algebra and user lattice manipulations while ensuring that the soundness, completeness and security that it originally guaranteed are not compromised. We also implement MLSK in a non-relational scenario, thereby demonstrating the extensibility of the model to other environments.