Some of the material in is restricted to members of the community. By logging in, you may be able to gain additional access to certain collections or items. If you have questions about access or logging in, please use the form on the Contact Page.
Memory hungry applications consistently keep their memory requirement curves ahead of the growth of DRAM capacity in modern computer systems. Such applications quickly start paging to swap space on the local disk, which brings down their performance, an old and ongoing battle between the disk and RAM in the memory hierarchy. This thesis presents a practical low-cost solution to this important performance problem. We give the design, implementation and evaluation of Anemone - an Adaptive NEtwork MemOry engiNE. Anemone pools together the memory resources of many machines in a clustered network of computers. It then presents an interface to client machines in order to use the collective memory pool in a virtualized manner, providing potentially unlimited amounts of memory to memory-hungry high-performance applications. Using real applications like the ns-2 simulator, the ray-tracing program POV-ray, and quicksort, disk-based page-fault latencies average 6.5 milliseconds whereas Anemone provides an average of latency of 700.2 microseconds, 9.2 times faster than using the disk. In contrast to the disk-based paging, our results indicate that Anemone reduces the execution time of single memory-bound processes by half. Additionally, Anemone reduces the execution times of multiple, concurrent memory-bound processes by a factor of 10 on the average. Another key advantage of Anemone is that this performance improvement is achieved with no modifications to the client's operating system nor the memory-bound applications due to the use of a novel NFS-based low-latency remote paging mechanism.
A Thesis submitted to the Department of Computer Science in partial fulfillment of the requirements for the degree of Master of Science.
Includes bibliographical references.
Kartik Gopalan, Professor Directing Thesis; Zhenhai Duan, Committee Member; An-i Andy Wang, Committee Member.
Florida State University
Use and Reproduction
This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s). The copyright in theses and dissertations completed at Florida State University is held by the students who author them.