Optional monitoring for long-lived transactions

Abstract

Runtime monitoring comes at a runtime cost. Overheads induced by monitoring and verification code may be necessary, and yet prohibitive in certain circumstances. When verification is local to a single unit of execution in a system, one can choose whether or not to monitor based on the risk of that individual unit. In this paper, we propose a monitoring and verification approach for a class of long-lived transaction-based systems whose execution can be partitioned into separate subtraces, one for each such transaction, and which are independent of each other from a correctness perspective. We focus on the use of this approach for the monitoring of smart contracts on distributed ledger technologies to show how we can reduce overheads in this manner.