Title

LSM-Forest: Planting Workload-optimized LSM-trees

DGIST Authors

Woojin Cho ; Sungjin Lee ; Yongjune Kim

Advisor

이성진

Co-Advisor(s)

Yongjune Kim

Issued Date

2022

Awarded Date

2022/02

Citation

Woojin Cho. (2022). LSM-Forest: Planting Workload-optimized LSM-trees. doi: 10.22677/thesis.200000593310

Type

Thesis

Subject

Big Data, LSM-tree, Database, Computer System, 빅데이터, 데이터베이스, 컴퓨터 시스템

Description

Big Data, LSM-tree, Database, Computer System, 빅데이터, 데이터베이스, 컴퓨터 시스템

Table Of Contents

I. Introduction 1
II. Background & Related Work 4
2.1 Background 4
2.1.1 LSM Tree Key-Value Store 4
2.1.2 Bloom Filter 5
2.1.3 MyRocks 5
2.1.4 Column Family & Multi Tenant 6
2.2 Related Work 7
2.2.1 Iso-KVSSD 7
2.2.2 Kevin 7
2.2.3 Parameter Tuning Using Deep Learning 8
III. Motivation 9
3.1 Problem of the Existing Shared Global LSM-tree 10
3.2 Problems with Existing Parameter Configuration 11
IV. LSM Forest 12
4.1 How to Planting LSM Tree 12
4.1.1 Target Application 13
4.1.2 Per Table Allocation LSM-tree 14
4.1.3 How to optimally allocate LSM-tree for each workload pattern 15
4.2 How to Gardening LSM Tree 16
4.2.1 Configure parameters according to workload characteristics 16
4.2.2 Resource allocate for each LSM-tree according to workload characteristics 19
V. Evaluation 27
5.1 Experimental Setup 27
5.1.1 Evaluation Setting 27
5.1.2 Comparison of each optimization 28
VI. Conclusions 30
References 31

URI

http://dgist.dcollection.net/common/orgView/200000593310
http://hdl.handle.net/20.500.11750/16301

DOI

10.22677/thesis.200000593310

Degree

Master

Department

Information and Communication Engineering

Publisher

DGIST

Show Full Item Record