In cellular networks, the densification of connected devices and base stations engender the ever-growing traffic intensity, and caching popular contents with smart management is a promising way to alleviate such consequences. Our research extends the previously proposed analysis of three-tier cache enabled Heterogeneous Networks (HetNets). The main contributions are threefold. We consider the more realistic assumption; that is, the distribution of small base stations is following Poisson-Poisson cluster processes, which reflects the real situations of geographic restriction, user dense areas, and coverage-holes. We propose the allocation of downlink data transmission capacity according to the cases of requested contents which are either cached or non-cached in nearby nodes and elucidate the traffic efficiency of the allocation under the effect of clustered deployment of small base stations. The throughput and delay of the allocation system are derived based on the approximated sojourn time of the Discriminatory Processor Sharing (DPS) queue. We present the results of achievable efficiency and such a system's performance for a better caching solution to the challenges of future cellular networks.