Abstract
According to the characteristics of data de-duplication, the metadata indexing structure based on B+ tree was introduced mainly for continuous data access applications, in which the metadata was divided into two kinds by popular and cold, with different layout strategies. As a consequence, the low power disks can stay in the standby state for a longer period of time with better energy efficiency, while effectively reducing random disk accesses with no additional hardware overhead. The method proposed in this paper makes it possible to store the index of the medium on disk only, rather than on disk and memory in previous studies.
| Original language | English |
|---|---|
| Pages (from-to) | 517-522 |
| Number of pages | 6 |
| Journal | Beijing Ligong Daxue Xuebao/Transaction of Beijing Institute of Technology |
| Volume | 34 |
| Issue number | 5 |
| Publication status | Published - May 2014 |
Keywords
- Data de-duplication
- Index structure
- Metadata
- Storage layout
