Unravelling the Impact of Temporal and Geographical Locality in Content Caching Systems

Abstract

To assess the performance of caching systems, the definition of a proper process describing the content requests generated by users is required. Starting from the analysis of traces of YouTube video requests collected inside operational networks, we identify the characteristics of real traffic that need to be represented and those that instead can be safely neglected. Based on our observations, we introduce a simple, parsimonious traffic model, named Shot Noise Model (SNM), that allows us to capture temporal and geographical locality of content popularity. The SNM is sufficiently simple to be effectively employed in both analytical and scalable simulative studies of caching systems. We demonstrate this by analytically characterizing the performance of the LRU caching policy under the SNM, for both a single cache and a network of caches. With respect to the standard Independent Reference Model (IRM), some paradigmatic shifts, concerning the impact of various traffic characteristics on cache performance, clearly emerge from our results.

A Decomposition Framework for Optimal Edge-Cache Leasing

“…However, convergence can be guaranteed for the Augmented Lagrangian (13) where λ := (λ s,f ), s ∈ S, f ∈ F is the dual vector with λ s,f ≥ 0 and ̺ > 0 is a penalizing factor. In order to maximize (13) in a distributed way, we apply Diagonal Quadratic Approximation (DQA) to find the primal solution for given dual values (see [1] for details). In every DQA iteration, a separate primal problem needs to be solved.…”

A Survey of Caching Networks in Content Oriented Networks

“…S. Travelso et al [43], [47] analyzed influence of temporal locality to cache performance by using trace-driven simulation. They compared cache size giving the same hit ratio for raw trace and shuffled trace (partially randomized trace).…”

Exploiting Path Diversity for Thwarting Pollution Attacks in Named Data Networking

“…Specifically, some popular objects may only be requested by users from one or a few locations and are not popular across the network. Geographical locality is observed between different countries and continents, where users have different languages and culture backgrounds, but is rarely observed within culturally homogeneous regions [26] [27]. For a PoP network that serves a limited geographical region (e.g., a university campus or a residential area), users tend to have highly homogeneous preferences and always prefer the same popular contents.…”

“…For a PoP network that serves a limited geographical region (e.g., a university campus or a residential area), users tend to have highly homogeneous preferences and always prefer the same popular contents. For example, a recent measurement study [27] on YouTube shows that the popular videos requested by users in one residential network are also popular in other residential networks. Because an NDN backbone router only caches the very popular data objects, we believe that there is no geographical locality for these objects within a PoP network.…”