FARMER: A novel approach to file access correlation mining and evaluation reference model |
| |
Authors: | Peng Xia Dan Feng and Fang Wang |
| |
Institution: | [1]Computer College, Huazhong University of Science and Technology, Wuhan 430074, P. R. China [2]Wuhan National Laboratory for Optoelectronic, Wuhan 430074, P. R. China |
| |
Abstract: | File semantic has proven effective in optimizing large scale distributed file system. As a consequence of the elaborate and
rich I/O interfaces between upper layer applications and file systems, file system can provide useful and insightful information
about semantic. Hence, file semantic mining has become an increasingly important practice in both engineering and research
community. Unfortunately, it is a challenge to exploit file semantic knowledge because a variety of factors could affect this
information exploration process. Even worse, the challenges are exacerbated due to the intricate interdependency between these
factors, and make it difficult to fully exploit the potentially important correlation among various semantic knowledges. This
article proposes a file access correlation miming and evaluation reference (FARMER) model, where file is treated as a multivariate
vector space, and each item within the vector corresponds a separate factor of the given file. The selection of factor depends
on the application, examples of factors are file path, creator and executing program. If one particular factor occurs in both
files, its value is non-zero. It is clear that the extent of inter-file relationships can be measured based on the likeness
of their factor values in the semantic vectors. Benefit from this model, FARMER represents files as structured vectors of
identifiers, and basic vector operations can be leveraged to quantify file correlation between two file vectors. FARMER model
leverages linear regression model to estimate the strength of the relationship between file correlation and a set of influencing
factors so that the “bad knowledge” can be filtered out. To demonstrate the ability of new FARMER model, FARMER is incorporated
into a real large-scale object-based storage system as a case study to dynamically infer file correlations. In addition FARMER-enabled
optimize service for metadata prefetching algorithm and object data layout algorithm is implemented. Experimental results
show that is FARMER-enabled prefetching algorithm is shown to reduce the metadata operations latency by approximately 30%–40%
when compared to a state-of-the-art metadata prefetching algorithm and a commonly used replacement policy. |
| |
Keywords: | storage management file correlation file system management mining method and algorithms |
本文献已被 维普 SpringerLink 等数据库收录! |
|