.png){kind=icon}
.png){kind=icon}
Return
Linked Block-based Multiversion B-Tree index for PCM-based embedded databases
.png){kind=icon}
PRE
AI
DOI：10.1016/j.sysarc.2015.08.002.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In this paper, by exploring the application characteristics of cyber-physical systems (CPS) and the performance characteristics of PCM, we propose a new B-tree index structure, called Linked Block-based Multi-Version B-Tree (LBMVBT), for indexing multi-version data in an embedded multi-version database for CPS. In LBMVBT, to reduce the number of writes to PCM in maintaining the index and improve the efficiency in serving version-range queries, we introduce the block-based scheme for indexing in which multiple versions of a data item are grouped into a version block to be indexed by a single entry in the multi-version index. To reduce the index re-organization cost (i.e., number of writes to PCM) due to node overflow and underflow, we add external entries in each index leaf node so that a node re-organization can be done by only updating pointers without copying the index entries of the re-organized leaf nodes to the new node. Analytic studies have been performed on LBMVBT and a series of experiments has been conducted to evaluate the efficacy of LBMVBT. The experimental results show that LBMVBT can effectively reduce the number of writes to the index and achieve a good overall performance on serving update transactions while version-range queries can be served with smaller number of reads to the index compared with MVBT. (C) 2015 Elsevier B.V. All rights reserved.
Keywords:
Phase change memory
Multi-version index
Cyber-physical systems
Embedded systems
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
4.1
Papers:
3.0K
Citations:
4.2K
Organization
View more organizations