13
Dilip N Simha, Maohua Lu, Tzi- cher chiueh An Update-Aware Storage System for Low-Locality Update-Intensive Workloads Park Chanhyun ASPLOS’12 March 3-7, 2012 London, England, UK.

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

  • Upload
    nasya

  • View
    64

  • Download
    0

Embed Size (px)

DESCRIPTION

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. - PowerPoint PPT Presentation

Citation preview

Page 1: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

Dilip N Simha, Maohua Lu, Tzi-cher chiueh

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

Park Chan-hyun

ASPLOS’12 March 3-7, 2012 London, England, UK.

Page 2: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

2

MotivationBackground BOSCExperimental MethodologyResults and analysisConclusion

Outline

Page 3: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

3

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

Motivation

Page 4: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

4

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 functionA 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)

New disk access interface

Page 5: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

5

Trail Disk Architecture

Background

Page 6: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

6

Background

Page 7: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

7

Batching Modifications with Sequen-tial Commit

Page 8: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

8

B tree B+ tree : file managementPort B+ tree index implementation using TPIE• Lock a leaf node before modifying it• Releases the lock after log update request

BOSC-Based B+ Tree

Page 9: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

9

Intel 2.4GHz CPU512KB L2 cache4GB memory400MH front-side busTwo Gigabit Ethernet interfaceFive 7200_RPM IBM Deskstar DTLA-307030 disks• 4:data disks• 1:logging disk

Evaluation Methodology

Page 10: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

10

Performance Improvement

Page 11: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

11

Sensitivity study

Page 12: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

12

Read query latency

Page 13: An Update-Aware Storage System for Low-Locality Update-Intensive Workloads

13

Solve problems of conventional storage system• An update-aware disk access interface• Efficient batched processing strategy

Deliver good performance with same durability guarantee

Conclusion