Request patterns and caching for VoD services with recommendation systems

Abstract

Abstract. Video on Demand (VoD) services like Netflix and YouTube account for ever increasing fractions of Internet traffic. It is estimated that this fraction will cross 80% in the next three years. Most popular VoD services have recommendation engines which recommend videos to users based on their viewing history, thus introducing time-correlation in user requests. Understanding and modeling this time-correlation in user requests is critical for network traffic engineering. The primary goal of this work is to use empirically observed properties of user requests to model the effect of recommendation engines on request patterns in VoD services. We propose a Markovian request model to capture the timecorrelation in user requests and show that our model is consistent with the observations of existing empirical studies. Most large-scale VoD services deliver content to users via a distributed network of servers as serving users requests via geographically co-located servers reduces latency and network bandwidth consumption. The content replication policy, i.e., determining which contents to cache on the servers is a key resource allocation problem for VoD services. Recent studies show that low start-up delay is a key Quality of Service (QoS) requirement of users of VoD services. This motivates the need to prefetch (fetch before contents are requested) and cache content likely to be requested in the near future. Since pre-fetching leads to an increase in the network bandwidth usage, we use our Markovian model to explore the trade-offs and feasibility of implementing recommendation based prefetching.

Vector Coded Caching Multiplicatively Boosts the Throughput of Realistic Downlink Systems

“…Example 1. Let us consider the Netflix library, focusing on movies, and let us make the educated speculation that the library entails a zipf parameter close to 1.4 (see for example [53]). Assume that we choose to apply coded caching on the part of the library that captures 90% of the traffic, such that on average, 90% of the Netflix traffic will experience a streaming volume reduction by a (theoretical) factor of G. Let us assume that the receiving devices are each endowed with a cache of size equal to 25GB, and let us assume that they stream HD movies whose size is approximately 1.3GB.…”

Modeling Request Patterns in VoD Services with Recommendation Systems

“…Modeling the request process: In a preliminary version of this work [11], we propose a Markovian model which captures the time-correlation in user requests in VoD services due to the presence of recommendation systems. We show that our model is consistent with empirically observed properties of request patterns in such VoD services [5,6,16,29].…”

Optimal Prefetching in Random Trees

“…In practical situations, the probabilities that the surfer moves to some new document after viewing the current document are not known a priori. However, these probabilities can be learned from data using Markov models, as in [6][7][8][9][10][11]. Moreover, the optimal control of a prefetching agent can be approximated using machine learning techniques such as reinforcement learning, as in [2].…”