Development towards 5G has introduced difficultchallenges in effectively managing and operating heterogeneousinfrastructures under highly varying network conditions. En-abling, for example, unified coordination and management ofradio resources across coexisting, multiple radio access technolo-gies (multi-RAT), require efficient representation using high-levelabstractions of the radio network performance and state. Withoutsuch abstractions, users and networks cannot harvest the fullpotential of increased resource density and connectivity optionsresulting in failure to meet the ambitions of 5G.We present a generic probabilistic approach for unified estima-tion of performance variability based on attainable throughputof UDP traffic in multi-RATs, and evaluate the applicability inan interface selection control case (involving WiFi and LTE)based on obtaining probabilistic user performance guarantees.From simulations we observe that both users and operators cansignificantly benefit from this improved service availability at lownetwork cost. Initial results indicate 1) 116% fewer performanceviolations and 2) 20% fewer performance violations with areduction by 35 times in the number of handovers, comparedto naive and state-of-the-art baselines, respectively.