Improving access to relevant data on faults, errors and failures in real systems

In order to be able to test the effectiveness and verify proposed techniques for enhanced availability based on field data from systems it is important to have reliability data of the components and the information necessary to characterize or model the system. This includes inter alia the type and number of components, their protection and dependency relations as well the automatic recovery mechanisms built into the system. An important benefit of making system models and logs available to the research community in a standard format is that it opens up the possibility for creating tools to assess and optimize deployed as well as hypothetical system configurations. Specialized tools for on-line and off-line analysis and classification of reliability data also become viable. Availability modeling tools could be benchmarked against actual data. Depending on the usefulness of such tools and the level of adoption of standard models and formats in the industry a market for reliability data analysis tools could emerge over time. These tools could be used during the design, deployment and operation phases of a system in order to predict or enhance the availability of the services it provides