Cost-Driven Data Caching in Edge-Based Content Delivery Networks

In this paper we study a data caching problem in edge-based content delivery network (CDN) where a data item is shared between service requests. Instead of improving the hit ratio with respect to limited capacity as in traditional case, we study the problem based on a semi-homogeneous (semi-homo) cost model in the edge-based CDN with monetary cost reduction as a goal. The cost model is semi-homo in the sense that all pairs of caching nodes have the same transfer cost, but each has its own caching cost rate. In particular, given a stream of requests
${\mathcal R}$
to a shared data item in the edge network, we present a shortest-path based optimal off-line caching algorithm that can minimize the total transfer and caching costs within
$O(mn)$
time (
$m:$
the number of network nodes,
$n:$
the length of request stream) in a proactive fashion. While for the online case, by extending the anticipatory caching idea, we also propose a 2-competitive online reactive caching algorithm and show its tightness by giving a lower bound of the competitive ratio as
$2-o(1)$
for any deterministic online algorithm. Finally, to combine the advantages of both algorithms and evaluate our findings, we also design a hybrid algorithm. Our trace-based empirical studies show that the proposed algorithms not only improve the previous results in both time complexity and competitive ratio, but also relax the cost model to semi-homogeneity, rendering the algorithms more practical in reality. We provably achieve these results with our deep insights into the problem and careful analysis of the solutions, together with a prototype framework.