This paper proposes an innovative data storage mechanism for decentralised workflow systems. The research is based on a peer-to-peer workflow architecture in which various nodes communicate with each other directly to carry out workflow functions. In this approach, a process is partitioned into separate tasks, and then individual tasks are passed to appropriate nodes. To realise this approach, a data repository independent process notation is provided. Moreover, this paper also designs communication protocols to maintain data in a dynamic environment. The key distinction of this approach is to propose a “know what you should know” policy to reflect the nature of applications.