Model Checking Inconsistency Recovery Costs

Abstract

Typically, when services become inconsistent from a business viewpoint, it is expected that compensation be used to recover from the inconsistency, by undoing the executed operations. In reality, compensation may incur additional costs, however existing approaches to recovery do not take such costs fully into account. We identify some major underlying gaps in SOC (Service Oriented Computing) related to compensation modelling, inconsistency identification, recovery and cost calculation. To make services more reasonable and predictable in dealing with inconsistencies from a cost perspective. We propose a cost-aware compensation framework modelling service compensations and compositions by a Petri Net-based model, reasoning about inconsistency recovery behaviours by model checking the LTL properties corresponding to business rules, and computing costs of recovery behaviours by parameterised cost calculation.

Towards cost-aware service recovery

“…Several Petri Net models [1,36] capture transactional workflows with preserving the true concurrency. Our previous work [44] extends those workflow nets to SOC domain by providing Net constructs for service composition modelling. However Petri Net-based service models are generally hard to analysis due to the large state space resulted from interleaving concurrent transitions.…”