On the optimization of storage capacity allocation for content distribution

Abstract

The addition of storage capacity in network nodes for the caching or replication of popular data objects results in reduced end-user delay, reduced network traffic, and improved scalability.The problem of allocating an available storage budget to the nodes of a hierarchical content distribution system is formulated; optimal algorithms, as well as fast/efficient heuristics, are developed for its solution. An innovative aspect of the presented approach is that it combines all relevant subproblems, concerning node locations, node sizes, and object placement, and solves them jointly in a single optimization step. The developed algorithms may be utilized in content distribution networks that employ either replication or caching/replacement. In addition to reducing the average fetch distance for the requested content, they also cater to load balancing and workload constraints on a given node. Strictly hierarchical, as well as hierarchical with peering, request routing models are considered.

Storage Capacity Allocation Algorithms for Hierarchical Content Distribution

“…The increase in the number of iteration has a somewhat larger impact on the required processing. The following proposition establishes an exact upper bound on the number of iterations performed by iGreedy (the proof is included in a longer version of this article Laoutaris et al, 2004b).…”

“…This allows calculating the gain for placing k in each of the n nodes in 0(n) instead of 0{n 2 ). In Laoutaris et al, 2004b we show how this can be achieved by first pre-computing information pertaining to request rates and closest parents and then using it to calculate up to n gain function for a given object in just 0(n). Since the gain function is 0(1) following the pre-computation step (occurring once at the beginning of each iteration), the complexity of each iteration of the Greedy algorithm depends on the number of nodes that are involved in the iteration and the update of the corresponding data structures.…”

“…At the beginning of each iteration there is the pre-processing step to get rate v {k) and par v (k) in 0(n) as explained in the appendix of Laoutaris et al, 2004b. The first step of each iteration requires that the highest value in all n max-heaps be selected.…”

Optimal content placement for peer-to-peer video-on-demand systems

“…Cache dimensioning problem is considered in [9], where Laoutaris et al optimized the storage capacity allocation for content distribution networks under a limited total cache storage budget, so as to reduce average fetch distance for the request contents with consideration of load balancing and workload constraints on a given node. Our paper takes a different perspective, focusing on many-user asymptotics so the results show that the finite storage capacity per node is never a bottleneck (even in the "large catalogue model", it also scales to infinity more slowly than the system size).…”

Local Utility Aware Content Replication

“…Joint formulations of the above mentioned problems have also appeared, e.g. in [6,7], where the proxy placement, proxy dimensioning, and object placement problems are combined into a single problem.…”