Epidemic-style diffusion schemes have been previously proposed for achieving peer-to-peer live streaming. Their performance trade-offs have been extensively studied for homogeneous systems, where all peers have the same upload capacity. However, epidemic schemes designed for heterogeneous systems have not been completely understood yet. In this paper we focus on the peer selection process and propose a generic model that encompasses a large class of algorithms. The process is modeled as a combination of two functions, an aware one and an agnostic one. By means of simulations, we analyze the awareness-agnostism trade-offs on the peer selection process and the impact of the source distribution policy in non-homogeneous networks. We highlight that a fairness trade-off arises between the performance of heterogeneous peers as a function of the level of awareness, and the strong impact the source selection policy and bandwidth provisioning have on diffusion performance.