Tycoon: A Distributed Market-based Resource Allocation System

TL;DR

The paper addresses resource allocation in large distributed P2P clusters where strategic users and dynamic demand complicate efficiency. It proposes Tycoon, a distributed market-based architecture that separates the allocation mechanism from user strategies, using Auction Share to schedule fine-grained resources with high utilization and low latency. Simulation results indicate market-based allocation maintains higher utility under strategic behavior and provides strong incentives for truth-telling, outperforming traditional proportional sharing. The authors plan kernel-level implementation and large-scale deployment to validate performance on realistic workloads, highlighting practical impact for Grid and PlanetLab-style systems.

Abstract

P2P clusters like the Grid and PlanetLab enable in principle the same statistical multiplexing efficiency gains for computing as the Internet provides for networking. The key unsolved problem is resource allocation. Existing solutions are not economically efficient and require high latency to acquire resources. We designed and implemented Tycoon, a market based distributed resource allocation system based on an Auction Share scheduling algorithm. Preliminary results show that Tycoon achieves low latency and high fairness while providing incentives for truth-telling on the part of strategic users.

Figures (1)

• Figure 1: The utility of different user behaviors and mechanism as a function of system load.