A transaction model defines the behaviour, constraints, integrity, inter-relationships, and robustness of database transactions.
Such models are generally evaluated indirectly, often by experiments on a database monitor that implements the model, or by
workload simulation. In this paper, we propose a novel method of comparing transaction models based on functions of architectural-
and isolation-work. Using these functions, we show the complexity of ten transaction models and discuss the relationship between
them. We conclude that our architectural- and isolation-work functions can be used to reason about transaction models and
as one measure for selecting the model appropriate to specific applications.