CloudBFT: Elastic Byzantine Fault Tolerance

Rodrigo Nogueira, Filipe Araujo, Raul Barbosa
2014 2014 IEEE 20th Pacific Rim International Symposium on Dependable Computing  
Cloud computing is increasingly important, with the industry moving towards outsourcing computational resources as a means to reduce investment and management costs, while improving dependability and performance. Nevertheless, the migration to cloud environments is a process that has been raising some concerns to many companies, which see the lack of physical control, the physical resources sharing (between distinct clients) and possible security breaches as the biggest barrier to move their
more » ... tems to cloud environments. Taking into account these cloud's problems and the resilience, availability and consistency needed in critical applications, we propose CloudBFT: a standard three-tiered system capable of taking advantage of cloud's scalability and elasticity, and simultaneously, being as resilient as possible in order to tolerate a wide range of faults, such as faults caused by intrusions, software and hardware faults, etc. The elasticity and scalability are achieved by scaling out and shrinking the cluster according to the processing requirements. On the other hand, to tolerate a wide range of faults (i.e., Byzantine faults), the system must execute a parallel version of a Byzantine fault-tolerant algorithm, where it is used groups of replicas placed on distinct physical machines, as a means to avoid exposing applications to correlated failures. This challenge becomes even more difficult in a relational model (as we used), where the synchronization and contention is higher. We believe that the elasticity we observe in our system, as it scales with the load, demonstrates the feasibility of tolerating Byzantine faults in a cloud-based web server using a relational data model. The results show that the system can scale with the load, as well as tolerating Byzantine faults in a cloud-based web server using a relational data model. Therefore, this work indicates that is possible to have a higher level of resilience in cloud environments and tolerating Byzantine faults without compromising the scalability and elasticity.
doi:10.1109/prdc.2014.31 dblp:conf/prdc/NogueiraAB14 fatcat:wiccrqbssjapthg22ey3aaidmy