hesela.devBORG block reorganization
BORG is a research block-layer system that adapts HDD data placement to observed access sequences.
A dedicated on-disk target holds selected copies; an indirection map tracks their current location. Its write buffer collects writes. [1]
ALIS instead evaluates heat and repeated-run clustering with trace-driven simulation; its design can keep multiple reorganized copies of a block. [2]
The author project page does not offer a BORG code release and points to EXCES as a reference, not the same implementation. [3]
Relations
- related
- DiskSeen, Table-based Prefetching (TaP), readahead (Linux syscall), Filesystem defragmentation
Limitations
- An unstable read working set can make BORG slower; reorganizing blocks also consumes I/O. [1]
Citations
- Bhadkamkar et al. BORG: Block-reORGanization for Self-optimizing Storage Systems. FAST 2009, sections 3-6 and 9.
- Hsu, Smith and Young. The Automatic Improvement of Locality in Storage Systems. UCB/CSD-03-1264, July 2003, sections 2-4 and 6.
- FIU Systems Research Laboratory: BORG project, Code and Traces.
Sources & caveats
- Historical HDD research; source synthesis, not a prototype reproduction or a modern SSD performance prediction.