"Design and performance evaluation of networked storage architectures" by Xubin He

Date of Award

2002

Degree Type

Dissertation

First Advisor

Qing Yang

Abstract

Data storage plays an essential role in today's fast-growing data-intensive network services. Within the past decades, processor speed and network bandwidth improve dramatically while disks have not kept pace with them in terms of access speed because of the mechanical nature of magnetic disks. This dissertation comprises of an intensive investigation on issues related to high performance networked storage architectures. First, this thesis proposes a novel networked storage architecture called SCSI-To-IP Cache Storage (STICS), which couples reliable and high-speed data caching with low-overhead conversion between SCSI and IP protocols. Through the efficient caching algorithm and localization of certain unnecessary protocol overheads, STICS can significantly improve performance, reliability, and scalability over current iSCSI systems. Numerical results using popular benchmarks such as VxBench, IoZone and PostMark, and EMC's trace have shown dramatic performance gain over the current iSCSI implementation. Second, this thesis introduces a peer-to-peer networked storage architecture called Distributed RAID and Location Independence Caching (DRALIC). The main idea of DRALIC is to combine or bridge the disk controller such as SCSI host bus adapter and network controller such as network interface card of PCs/Workstations interconnected by a high-speed LAN switch. Our preliminary study has shown that this architecture has significant advantages for web server applications in terms of performance and reliability. Third, this thesis proposes a software RAID with a large virtual NVRAM cache (VC-RAID), which provides high performance and reliability for networked storage. VC-RAID uses a combination of a small portion of the system RAM and a log disk to form a hierarchical cache. Our measurements on a prototype VC-RAID show that the VC-RAID has superb performance advantages over the built-in software RAID shipped with the Linux package. Fourth, this thesis presents an intensive study on the performance of distributed web server architectures under realistic workloads with a mixture of static web page requests, CGI requests, Servlet requests, and database queries. We observed that performance behaviors of the web server architectures considered under this mixed workload are quite different from that under static page workload and sometimes counter-intuitive.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.