arXiv:1501.03446v1 [cs.NI] 14 Jan 2015 Flexible Content Placement in Cache Networks using Reinforced Counters Guilherme Domingues 1 , Edmundo de Souza e Silva 1 , Rosa M. M. Leao 1 , Daniel S. Menasche 1 1 Universidade Federal do Rio de Janeiro (UFRJ) Rio de Janeiro, RJ – Brazil {guilhdom, edmundo, rosam, sadoc}@land.ufrj.br Abstract. In this paper we study the problem of content placement in a cache network. We consider a network where routing of requests is based on random walks. Content placement is done using a novel mechanism referred to as “rein- forced counters”. To each content we associate a counter, which is incremented every time the content is requested, and which is decremented at a fixed rate. We model and analyse this mechanism, tuning its parameters so as to achieve desired performance goals for a single cache or for a cache network. We also show that the optimal static content placement, without reinforced counters, is NP hard under different design goals. 1. Introduction In today’s Internet the demand for multimedia files and the sizes of these files are steadily increasing. The popularity of Youtube, Dropbox and one-click file sharing systems, such as Megaupload, motivate researchers to seek for novel cost-effective content dissemina- tion solutions. Caching is one of the most classical solutions to increase the load supported by computer systems. In essence, caching consists of transparently storing data in a way that future requests can be served faster. Caching in the realm of standalone computer architectures received significant attention from the research community since the early sixties, and web-caching has also been studied for at least one decade. The objects of study of this work, in contrast, are cache networks, which were proposed and started to receive focus much more recently [Zhang and Carofiglio 2012, Jacobson et al. 2009]. Cache networks comprise two main features: caching and routing, both performed by the same core component: a cache router. In a cache network, content traverses the network from sources to destinations, but can in turn be stored in caches strategically pla- ced on top of the routers. The data plane is responsible for transmitting contents whereas the control plane is responsible for the routing of requests. The system considered in this paper is illustrated in Figure 1. A request for file F is routed using random walks through the caches. Every time the request hits a cache, the cache selects uniformly at random one of the links that are incident to the cache and uses it to forward the request. Once the request reaches a cache where the content is stored, the content is transferred to the requester through the data plane. We assume that all contents are stored in at least one cache. In this paper we study the problem of content placement in cache networks. We pose the following questions: (1) How to store and evict contents from a cache in a