1 / 13

An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

An Update-Aware Storage System for Low-Locality Update-Intensive Workloads. Dilip N Simha , Maohua Lu, Tzi-cher chiueh. ASPLOS’12 March 3-7, 2012 London, England, UK. Park Chanhyun. Outline. Motivation Background BOSC Experimental Methodology Results and analysis Conclusion.

nasya
Download Presentation

An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. An Update-Aware Storage System for Low-Locality Update-Intensive Workloads Dilip N Simha, Maohua Lu, Tzi-cherchiueh ASPLOS’12 March 3-7, 2012 London, England, UK. Park Chanhyun

  2. Outline • Motivation • Background • BOSC • Experimental Methodology • Results and analysis • Conclusion

  3. Motivation • I/O access delay • Seek rotational, transfer • Low-locality, update-intensive disk access workload • Disk buffering, caching, scheduling • Simple read/write interfaces are not adequate • Read access has higher priority. • Update requests from many applications

  4. New disk access interface • Conventional disk access interface • Read -> modify -> write • Read(target_block_addr, dest_buf_addr) • Write(target_block_addr, src_buf_addr) • Allow applications of a storage system • Disk access request as an update. • Associate with an update request, a callback function • A new storage system architecture : BOSC • Batching mOdifications with Sequential Commit • Between storage applications and hardware storage system • Modify(target_block_addr, ptr_modification, ptr_commit_function)

  5. Background • Trail Disk Architecture

  6. Background

  7. Batching Modifications with Sequential Commit

  8. BOSC-Based B+ Tree • B tree B+ tree : file management • Port B+ tree index implementation using TPIE • Lock a leaf node before modifying it • Releases the lock after log update request

  9. Evaluation Methodology • Intel 2.4GHz CPU • 512KB L2 cache • 4GB memory • 400MH front-side bus • Two Gigabit Ethernet interface • Five 7200_RPM IBM Deskstar DTLA-307030 disks • 4:data disks • 1:logging disk

  10. Performance Improvement

  11. Sensitivity study

  12. Read query latency

  13. Conclusion • Solve problems of conventional storage system • An update-aware disk access interface • Efficient batched processing strategy • Deliver good performance with same durability guarantee

More Related