CycloidGrid: A proximity-aware P2P-based resource discovery architecture in volunteer computing systems

Abstract

Volunteer computing which benefits from idle cycles of volunteer resources over the Internet can integrate the power of hundreds to thousands of resources to achieve high computing power. In such an environment the resources are heterogeneous in terms of CPU speed, RAM, disk capacity, and network bandwidth. So finding a suitable resource to run a particular job becomes difficult. Resource discovery architecture is a key factor for overall performance of peer-to-peer based volunteer computing systems. The main contribution of this paper is to develop a proximity-aware resource discovery architecture for peer-to-peer based volunteer computing systems. The proposed resource discovery algorithm consists of two stages. In the first stage, it selects resources based on the requested quality of service and current load of peers. In the second stage, a resource with higher priority to communication delay is selected among the discovered resources. Communication delay between two peers is computed by a network model based on queuing theory, taking into account the background traffic of the Internet. Simulation results show that the proposed resource discovery algorithm improves the response time of user’s requests by a factor of 4.04 under a moderate load.

Introduction

Volunteer computing in which volunteers donate processing and storage resources is an attractive cost efficient platform for running scientific projects with significant computational requirements (tens or hundreds of TeraFLOPs) [1]. Some of these projects are the SETI@home [2], Folding@home [3], EDGeS [4], DEGISCO [5] and EDGI [6] and Climate@home [7] projects.

Some of the popular volunteer computing systems are BOINC [8], [9], condor-like grid system [10], Entropia [11], XtremeWeb [12], Aneka [13], and SZTAKI [14]. Peer-to-Peer (P2P) based volunteer computing (VC) systems represent a decentralized, self-organized and scalable environment for running applications such as PastryGrid [15], BonjourGrid [16], ShareGrid [17], Condor-Flock P2P [18], and Self-Gridron [19].

A fundamental challenge in this large, decentralized and distributed resource sharing environment is efficient discovery of resources described by a set of attributes such as CPU speed, memory, and operating system type. Another challenge comes from these complex environments characterized by large scale and geographically scattered resources. In such an environment, millions of heterogeneous resources are scattered across geographically distributed nodes, therefore a resource discovery architecture with proximity-aware features has great impact on overall performance of the system [18]. There are many resource discovery algorithms in P2P-based volunteer computing systems [20], [21], [22], [23], [24], but none of them considers a model in computing communication overhead.

The main contribution of this work is to propose a proximity-aware resource discovery architecture considering quality of service (QoS) constraints of requests, and communication overhead in P2P-based VC systems. In a previous work [25], we presented an architecture for resource discovery in P2P-based VC systems. This architecture is called CycloidGrid, and it distributes a load between peers based on QoS constraints of requests, round trip time (RTT), and current load of resources (for more information refers to Section 3). In this research we focus on a network model based on queuing theory to compute more realistic communication overhead considering the background traffic of the Internet.

The proposed resource discovery algorithm is composed of two stages. In the first stage, resources are selected based on QoS constraints of requests and current load of peers. The QoS constraints can be selected from CPU speed, RAM requirements, hard disk requirements, operating system, and processor model. In the second stage, the algorithm selects a resource with higher priority to communication delay among the discovered resources in the first stage. To compute the communication delay between two peers in the system, each connection is modeled as a GI/GI/1 queue. Background traffic of the Internet is considered in the model to generate a more realistic communication delay. In summary, we have the following contributions in this paper:

• We provide a resource discovery architecture in P2P-based volunteer computing systems considering QoS constraints of request and communication overhead.

• We adapt the analytical model for computing communication delay between two peers. The background traffic of the Internet is considered during computation of communication delay between two peers.

• We evaluate the proposed resource discovery architecture under realistic workload models and different numbers of peers to show the scalability of the system.

The rest of this paper is organized as follows: Section 2 presents related work. Section 3 discusses the CycloidGrid environment, including the application model, architecture, and churn management. Section 4 presents an analytical model for computing communication delay in CycloidGrid and resource discovery policy in this architecture. Section 5 describes the performance evaluation of the proposed resource discovery architecture under a realistic workload model. Conclusions and future directions are presented in Section 6.