[Paper Review] Optimal Content Placement for Peer-to-Peer Video-on-Demand Systems

Bo Bo, Tân+1 more|arXiv (Cornell University)|Apr 27, 2010

Caching and Content Delivery17 references4 citations

TL;DR

This paper proposes asymptotically optimal content placement strategies for peer-to-peer video-on-demand systems to maximize uplink bandwidth utilization. It introduces a 'proportional-to-product' placement and counter-based request acceptance rule, proving that bandwidth loss vanishes as peer storage capacity grows unboundedly, even arbitrarily slowly, under large catalogue scaling.

ABSTRACT

In this paper, we address the problem of content placement in peer-to-peer systems, with the objective of maximizing the utilization of peers' uplink bandwidth resources. We consider system performance under a many-user asymptotic. We distinguish two scenarios, namely "Distributed Server Networks" (DSN) for which requests are exogenous to the system, and "Pure P2P Networks" (PP2PN) for which requests emanate from the peers themselves. For both scenarios, we consider a loss network model of performance, and determine asymptotically optimal content placement strategies in the case of a limited content catalogue. We then turn to an alternative "large catalogue" scaling where the catalogue size scales with the peer population. Under this scaling, we establish that storage space per peer must necessarily grow unboundedly if bandwidth utilization is to be maximized. Relating the system performance to properties of a specific random graph model, we then identify a content placement strategy and a request acceptance policy which jointly maximize bandwidth utilization, provided storage space per peer grows unboundedly, although arbitrarily slowly, with system size.

Motivation & Objective

• To maximize utilization of peers’ uplink bandwidth in P2P video-on-demand systems by optimizing content placement.
• To address two system architectures: Distributed Server Networks (DSN) with external requests and Pure P2P Networks (PP2PN) with internal requests.
• To analyze performance under both limited and large content catalogue scaling, particularly focusing on asymptotic optimality.
• To establish that unbounded growth in per-peer storage is necessary for maximal bandwidth utilization in large catalogue settings.
• To design a request acceptance policy that, when paired with optimal placement, ensures vanishing loss probability as system size increases.

Proposed method

• Models the system as a loss network, where requests are rejected if they would disrupt ongoing services.
• Proposes a 'proportional-to-product' placement strategy that replicates content in proportion to its popularity and the number of potential requesters.
• Introduces a counter-based acceptance rule: each request is associated with L randomly selected peers holding the content, and is rejected if any peer’s counter exceeds a threshold.
• Uses a random graph model to relate system performance to structural properties of peer-content connectivity.
• Applies large-system asymptotic analysis to derive conditions under which bandwidth utilization approaches optimality.
• Imposes a technical assumption that contents must be replicated at least $ M^{3/4} $ times to be eligible for service.

Experimental results

Research questions

• RQ1 What content placement strategy maximizes uplink bandwidth utilization in a P2P VoD system under a loss network model?
• RQ2 How does the optimality of content placement depend on system architecture—specifically, DSN versus PP2PN?
• RQ3 What scaling behavior of the content catalogue size is required to achieve optimal bandwidth utilization?
• RQ4 Can a simple request acceptance policy, combined with optimal placement, ensure vanishing request rejection as system size grows?
• RQ5 Is there a minimal storage requirement per peer for achieving optimal performance, or does any unbounded growth suffice?

Key findings

• The 'proportional-to-product' placement strategy is asymptotically optimal for Distributed Server Networks (DSN) under limited catalogue size.
• For Pure P2P Networks (PP2PN), a distinct 'Hot-Warm-Cold' placement strategy is proven optimal in the limited catalogue regime.
• In the large catalogue scaling regime, the 'proportional-to-product' strategy remains optimal when paired with a modified request management technique.
• Storage space per peer must grow unboundedly, although arbitrarily slowly, to achieve maximal bandwidth utilization in the large catalogue setting.
• With the counter-based acceptance rule and optimal placement, the probability of request rejection, $ ho(M) $, decreases to zero as $ M \to \infty $, ensuring near-perfect bandwidth utilization.
• The system achieves optimal performance without requiring a minimal growth rate of storage (e.g., no $ \log B $ lower bound is needed).