Storage, Communication, and Load Balancing Trade-off in Distributed Cache Networks

Abstract

Abstract-We consider load balancing in a network of caching servers delivering contents to end users. Randomized load balancing via the so-called power of two choices is a wellknown approach in parallel and distributed systems. In this framework, we investigate the tension between storage resources, communication cost, and load balancing performance. To this end, we propose a randomized load balancing scheme which simultaneously considers cache size limitation and proximity in the server redirection process.In contrast to the classical power of two choices setup, since the memory limitation and the proximity constraint cause correlation in the server selection process, we may not benefit from the power of two choices. However, we prove that in certain regimes of problem parameters, our scheme results in the maximum load of order Θ(log log n) (here n is the network size). This is an exponential improvement compared to the scheme which assigns each request to the nearest available replica. Interestingly, the extra communication cost incurred by our proposed scheme, compared to the nearest replica strategy, is small. Furthermore, our extensive simulations show that the trade-off trend does not depend on the network topology and library popularity profile details.

Balanced allocation on graphs: A random walk approach

“…where Empty t (T j ) is the size of empty nodes contained in T j after allocating t balls. In what follows we find a lower bound for (15). Let  t+1 denote the event that the number of nonempty nodes in N r (u j−1 ) after allocating t balls is at most ( − 1) r −1 ∕2.…”

“…In particular, they showed that the maximum number of visits to a node is at most (1+o(1))⋅ log n log log n . Our result can also be seen as an application of the mathematical concept of NBRWs to the task allocation in distributed networks, where the goal is to minimize the maximum load imposed on each server (for a similar model, e.g., see [15]). In such a model, a set of processors are connected over a network and the processors randomly receive requests in sequential rounds.…”

Distribuição de Conteúdo Sob Demanda Através da Alocação de Microserviços Dinâmicos na Borda e Núcleo da Rede

“…Tang, Tay e Wen exploram o redirecionamento de requisições para vários servidores buscando minimizar uma função de custo baseada em tempo de resposta, custos computacionais e capacidade da rede [Tang et al 2014]. Siavoshani, Pourmiri e Shariatpanahi propõem o balanceamento de recursos armazenamento e comunicação utilizando o redirecionamento de requisições baseado em limitações no espaço de cache e proximidade de clientes ao servidor [Siavoshani et al 2017]. Chunlin, Chuanli e Youlong buscam minimizar o tempo de execução e consumo de energia de serviços de distribuição de vídeo através do uso de parâmetros de QoS para a seleção de serviços [Chunlin et al 2019].…”

“…Características QoE QoS Cache Requisição Computação em névoa DFA DRA [Tang et al 2014] [Siavoshani et al 2017] [Chunlin et al 2019] [Xiao et al 2015] [Zhang et al 2016] [Aazam and Huh 2015] [Aazam et al 2016] [Ni et al 2017 Com base na análise dos trabalhos relacionadosé possível concluir que serviços de VoD implantados em ambientes de computação em névoa com múltiplos níveis exploram a unificação de recursos de diferentes dispositivos de rede a fim de ampliar sua QoE. Portanto, torna-se essencial direcionar as requisições para nós de computação em névoa adequados, assim evitando que usuários abandonem o serviço devido a uma baixa QoE.…”

Autonomous allocation of wireless body-sensor networks to frequency channels: modeling with repeated “balls-in-bins” experiments

“…In particular, results are available for the distributions of the minimum and maximum occupancy of an urn [14]. They also have found applications in wireless channel allocation problems [7], [1], load balancing problems for cache servers or cloud computing in the Internet [37], [35], [3], in task allocation problems in crowdsourcing applications [23], or in the analysis of hashing schemes [11]. In many of these papers an assumption is made that a "ball" looking for a "bin" has a chance to inspect the contents of a number of randomly chosen bins before making any decision, and then it will pick the most favorable (e.g.…”