2016/09/08 by Samarth Gupta, Gupta, Samarth, Sharayu Moharir +1
Computer Science · #Caching and Content Delivery #FOS: Computer and information sciences #Network Traffic and Congestion Control #Networking and Internet Architecture (cs.NI) #Peer-to-Peer Network Technologies
paper · pdf · doi:10.48550/arxiv.1609.02391
openalex publication_date 2016/09/08 · openalex created_date 2022/10/03 · openalex updated_date 2026/07/28
Video on Demand (VoD) services like Netflix and YouTube account for ever\nincreasing fractions of Internet traffic. It is estimated that this fraction\nwill cross 80% in the next three years. Most popular VoD services have\nrecommendation engines which recommend videos to users based on their viewing\nhistory, thus introducing time-correlation in user requests. Understanding and\nmodeling this time-correlation in user requests is critical for network traffic\nengineering. The primary goal of this work is to use empirically observed\nproperties of user requests to model the effect of recommendation engines on\nthe request patterns in VoD services. We propose a Markovian request model to\ncapture the time-correlation in user requests and show that our model is\nconsistent with the observations of existing empirical studies.\n Most large-scale VoD services deliver content to users via a distributed\nnetwork of servers as serving users requests via geographically co-located\nservers reduces latency and network bandwidth consumption. The content\nreplication policy, i.e., determining which contents to cache on the servers is\na key resource allocation problem for VoD services. Recent studies show that\nlow start-up delay is a key Quality of Service (QoS) requirement of users of\nVoD services. This motivates the need to pre-fetch (fetch before contents are\nrequested) and cache content likely to be request in the near future. Since\npre-fetching leads to an increase in the network bandwidth usage, we use our\nMarkovian model to explore the trade-offs and feasibility of implementing\nrecommendation based pre-fetching.\n