Home

The goal of this work is to design and implement deployment-ready technologies for generic services to be deployed on serverless cloud platforms, while making good use of the available resources to achieve a given performance goal but also aligned with cost considerations. As a demonstrating case study we will use a Byzantine fault tolerant state machine replication service in the context of a validator network, which should provide good performance even under faults. One of the grantees will focus on developing a shim layer for the deployment of serverless functions, so that a good use of resources is achieved by making an informed runtime decision on whether to spawn a new serverless instance or attach the new function to a running instance as an additional thread. In addition to the design and implementation of this shim layer, the grantee will conduct an experimental evaluation of the system. The other grantee will focus on developing an efficient state machine replication service that tolerates Byzantine faults in the context of a validator network, as a case study demonstrating a service that can be deployed efficiently on the cloud or other types of execution environments. For increased efficiency, this state machine replication service should make use of our recently developed asynchronous protocol. The grantee will be responsible for the implementation and experimental evaluation of the implemented prototype.