| Carl-Friedrich-Gauß-Fakultät | Informatik

Self-adaptive partitioning of state service in Byzantine Fault Tolerance system

Betreuer Bijun Li
Wenbo Xu
Professor Prof. Dr. Rüdiger Kapitza
Projekt railcloud
IBR Gruppe DS (Prof. Kapitza)
Art Projektarbeit
Status laufend


The traditional Byzantine Fault Tolerance (BFT) state machine replication (SMR) system requires a total order of all requests. There is already new mechanism exploiting parallelism in modern replication systems. The entire state of a service can be split into different partitions and therefore requires only a partial order on the requests visiting the same partition. Our previous work [1] describes such a multi-leader BFT framework. This however arises the new challenge to define the partitions in a proper way, so that the requests crossing multiple partitions can be minimized.


A previous work [2] has explored the possibility of using Machine Learning techniques to smartly partition the service state objects, according to the data access dependency of client requests. It has proposed a combination of linear regression and recursive bi-partitioning approach to learn the underlying hardware behavior and the knowledge about application requests, and eventually to partition the system state into clusters of highly dependent objects that are represented in undirected weighted graphs. The goal of this work is to complete the learning approach so that it can be used by the BFT system efficiently. It covers two aspects: 1) Validate and improve the performance of learning approach 2) Integrate the learning approach into the BFT system to form a self-adaptive partitioning method. A very good skill of Java and basic machine learning knowledge are required.

[1]: https://www.ibr.cs.tu-bs.de/theses/bli/ml-bft.html

[2]: https://www.ibr.cs.tu-bs.de/theses/bli/partitioning.html

aktualisiert am 06.06.2017, 13:17 von Bijun Li