Submitted Abstract
One of the major stepping stones for a wide application of fault and intrusion tolerance techniques, such as state-machine replication (BFT-SMR), lies in the overheads of these solutions in terms of the number of replicas required. Although hybridization allowed cutting the replication degree (e.g., from 3f+1 to 2f+1 replicas when tolerating f faults) and although reactive protocols reduced the number of active replicas by an additional f (to be reactivated for error recovery), a fundamental limitation remains. The fault threshold f must be chosen at time of deployment and remains fixed over the lifetime of the system. The goal of this joint FNR-FCT research project is to explore methods and protocols for dynamically adjusting the set of replicas required in BFT SMR protocols. Fault adaptivity will allow adjusting the replication degree to the threat level assessed by a risk managing distributed control plane, operating across multiple domains. Location adaptivity allows replicas to securely follow load peaks. In ThreatAdapt, we will study fault, location and combined fault-and-location adaptivity with homogeneous and hybrid system models, evaluate the developed protocols through simulation and over the internet and apply the lessons learned from generic BFT-SMR protocols on the example of a fault-adaptive blockchain.