Thesis
A cost-driven replacement policy for hierarchical key-value store
Washington State University
Master of Science (MS), Washington State University
2012
Handle:
https://hdl.handle.net/2376/102157
Abstract
In recent years, cloud computing has been used to support a plethora of applications ranging from simple websites to giant, data- and compute-driven services. Previously we have developed a caching system designed to utilize the elasticity of the cloud in an automated cost-aware approach. In this thesis we continue exploration of our elastic key-value cache as a means for accelerating both data- and compute-intensive applications. We present a hybridized cloud caching system that utilizes in-instance memory, in-instance disk, and a persistent storage medium. Further, we create a hierarchy from these mediums to reflect the importance of the data within, and provide mechanisms for placing this data within the hierarchy. Utilizing Amazon Web Services (AWS), we construct experiments designed to evaluate a number of data eviction and data placement schemes. As is common among web applications, we configure our cache to exist on an edge network with a remote back-end data store, introducing a degree of added latency for cache misses. The data store consists of a number of files in a size and query pattern representative of the average web application. These experiments measure query latency and in which medium of the storage hierarchy cache hits occur. We show that our cost aware approach to data placement demonstrates approximately a 10% improvement in query latency over Least Recently Used (LRU) and a significantly higher number of hits in memory.
Metrics
12 File views/ downloads
11 Record Views
Details
- Title
- A cost-driven replacement policy for hierarchical key-value store
- Creators
- Travis Hall
- Contributors
- David Chiu (Degree Supervisor)
- Awarding Institution
- Washington State University
- Academic Unit
- Electrical Engineering and Computer Science, School of
- Theses and Dissertations
- Master of Science (MS), Washington State University
- Publisher
- Washington State University; Pullman, Wash. :
- Identifiers
- 99900525009301842
- Language
- English
- Resource Type
- Thesis