This paper aims at presenting how a number of recent modeling approaches can be used for better understanding microbial ecosystems dynamics. In first part, an important question – the ability of certain ecosystems to exhibit overyielding – is investigated using a model-based approach. It is shown that classical competition theory cannot explain such phenomenon, thus invalidating a large class of classical mass-balance-based models, Rapaport et al. (2019). In second part, we show how new combinatorial approaches can be used to find the best combination of species of a functional ecosystem with limited complexity. More precisely classification approaches inspired from the work by Jaillard et al. (2018) are used and illustrated with simulations. Their robustness with respect to a number of experimental parameters (investigated in simulation) is studied. For ecosystems with higher richness, we show how another probabilistic approach proposed by Jaillard et al. (2014) may be useful. Keywords: Modeling, microbial ecosystems, microbial interactions, mathematical ecology, diversity, community assemblage.