2022 Fiscal Year Final Research Report
Adaptive Replica Relocation of Geo-replicated State Machines on Public Cloud
| Project/Area Number |
18K18029
|
|---|
| Research Category |
Grant-in-Aid for Early-Career Scientists
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Basic Section 60050:Software-related
|
|---|
| Research Institution | Toyohashi University of Technology |
|---|
Principal Investigator |
Nakamura Junya 豊橋技術科学大学, 情報メディア基盤センター, 准教授 (60739746)
|
|---|
| Project Period (FY) |
2018-04-01 – 2023-03-31
|
| Keywords | BFTレプリケーション / ビザンチン故障 / 地理的分散 / パブリッククラウド / ビザンチン合意 |
|---|
| Outline of Final Research Achievements |
Geographical BFT replication replicates a service, which is defined as a state machine, to multiple geographically distributed replicas. The replicas work in cooperation with each other and provide Byzantine fault tolerance to the service. In this research project, we developed the following two methods to enable efficient geographically distributed BFT replication that can respond to dynamic changes in the replication environment. The first method determines replica deployment that achieves optimal latency under given conditions. The second method efficiently transfers service state to remote replicas.
|
| Free Research Field |
分散システム
|
| Academic Significance and Societal Importance of the Research Achievements |
地理的分散BFTレプリケーションはパブリッククラウドの発達によって容易に実現できるようになったが，レプリカ間の通信帯域の変動などの影響を受けて，その性能は常に変化する．良好なレプリケーション性能を維持するためには，定期的に最適なレプリカ配置を計算し，計算結果に基づいてレプリカを移動させる必要があるが，これらの問題は既存研究ではほとんど考慮されてこなかった．本研究では新たに2つの手法を考案して問題を解決し，地理的分散BFTレプリケーションの実用性を向上した．
|
