Modelling study

A model will be created to simulate parts of the inputs, inner workings and output of the IceCube reconciliation algorithm. The algorithm is based upon the concept of logs, which are sequences of actions. In this context, an action is a modification made to a local replica of a global state, which includes an operation and some context-dependent data. The basis of the model is the combined set of actions from the logs to be reconciled.

Visualising this algorithm in its general form is not feasible, even for a fixed number of actions. The main reason for this is that constraints have to be specified by the application programmer on the order of actions, and these constraints depend on the object which the action influences. In addition to considering only a limited number of actions, the number of objects (action targets) and the types of action will also be restricted. The usefulness of the model will not be compromised, however, as even logs which are limited in size can give rise to quite complex interactions in the reconciliation engine when combined with other logs of similar size.

What the model aims to illustrate and simulate is how IceCube’s reconciliation algorithm works. The goal of the development is to make the model as interactive as possible, that is, the user should be able to influence the inputs to the algorithm at any point as well as clearly see the results of his actions. For an application programmer, one possible use of the model is to specify the constraints between actions (which are input to the algorithm) in such a way that the algorithm produces its output in as few steps as possible, while still maintaining correctness conditions.

This modelling study requires in-depth understanding of the IceCube technology. It is therefore imperative that the developer has a thorough background in this research. As the purpose of the model lies within the field of educational technology, sources in this area will also have to be consulted.