CSR: EAGER: A Scalable and Efficient Resource Discovery System for Large-Scale Distributed Systems

CSR：EAGER：适用于大型分布式系统的可扩展且高效的资源发现系统

• 批准号: 1249603
• 负责人: Haiying Shen
• 金额: $ 20万
• 依托单位: Clemson University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2012
• 资助国家: 美国
• 起止时间: 2012-09-01 至 2015-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1249603&HistoricalAwards=false
• 关键词: CSR; EAGER; Scalable; Efficient; Resource

The objective of this EAGER project is to explore a preliminary design of a Scalable and Efficient Resource Discovery system (SERD) that can be utilized in large scale distributed systems. In particular, the PI is investigating the incorporation of the following three key innovations into SERD.(1) A scalable middleware for resource discovery:Current resource discovery systems use flat peer-to-peer (P2P) structures that lead to either high overhead in structure maintenance or low efficiency in resource discovery. They also lack proximity-awareness and multi-resource range querying capabilities. By contrast, SERD clusters proximity-close nodes that have the same resource and connects the clusters using a backbone hierarchical P2P structure, permitting efficient range querying of proximity-close resources.(2) Scalable and efficient resource discovery: Using the SERD middleware in (1), the PI will develop efficient resource information marshaling and searching algorithms. These algorithms can achieve proximity awareness and can provide range querying capabilities that enables the nodes to find physically close resources within specified ranges. (3) Balanced resource discovery:Current resource discovery systems lack a load balance algorithm to balance the resource discovery load among the different nodes. By contrast, SERD incorporates a load balancing algorithm in order to balance the resource discovery load among the different nodes.

EAGER项目的目标是探索一个可扩展的和有效的资源发现系统（SERD），可用于大规模分布式系统的初步设计。特别是，PI正在研究将以下三个关键创新纳入SERD。(1)一个可扩展的资源发现中间件：目前的资源发现系统使用扁平的对等（P2P）结构，导致结构维护开销高或资源发现效率低。它们还缺乏邻近感知和多资源范围查询能力。相比之下，SERD集群的邻近接近节点，具有相同的资源，并连接集群使用骨干分层P2P结构，允许有效的范围查询邻近接近的资源。(2)可伸缩和高效的资源发现：使用（1）中的SERD中间件，PI将开发高效的资源信息编组和搜索算法。这些算法可以实现邻近感知，并且可以提供范围查询能力，该范围查询能力使得节点能够在指定范围内找到物理上接近的资源。(3)均衡的资源发现：目前的资源发现系统缺乏负载均衡算法来平衡不同节点之间的资源发现负载。相比之下，SERD结合了负载平衡算法，以平衡不同节点之间的资源发现负载。