Consistency Maintenance using UNIFY

Authors:Natrajan, Anand, Department of Computer ScienceUniversity of Virginia Reynolds, Jr, Department of Computer ScienceUniversity of Virginia Srinivasan, Sudhir, Department of Computer ScienceUniversity of Virginia

Distributed simulations comprised of aggregated entities (AEs) and disaggregated entities (DEs) pose critical consistency issues when AEs and DEs interact. Usually, meaningful interaction cannot take place without one of the two representing itself at a level of resolution compatible with the level of the other. Any approach that employs dynamic transitions between aggregated and disaggregated resolution levels suffers from not only potential consistency problems, but also chain disaggregation, network flooding, transition latency, and mapping problems between levels. Solutions meant to solve some or all of these problems leave one critical issue unresolved: proper and efficient maintenance of consistency among the levels of resolution for the same set of objects. An alternative approach, where AE and DE levels of resolution are maintained concurrently for the same objects, encounters the same consistency issues: anything that happens to a DE must be reflected accurately in the AE, and vice versa. The fundamental problem is this: consistency cannot be guaranteed without a unified, coherent approach to correct, efficient consistency maintenance among levels of resolution for a set of simulated objects. In our approach, UNIFY, we contend that multiple levels of resolution should be addressed inside individual simulations, in order to ensure an efficient, coherent, verifiable solution. We propose the concept of MREs (Multiple Resolution Entities) in lieu of AEs and DEs. MREs are capable of representing simulated objects at specified levels of resolution in a consistent manner. A MRE should be able to provide, when requested, bindings for attributes at any desired level in a timely manner. A number of key issues must be addressed in order to provide this capability. The most critical issues are: identification of core data, temporal consistency, and mapping consistency.

All rights reserved (no additional license for public reuse)
Source Citation:

Natrajan, Anand, Jr Reynolds, and Sudhir Srinivasan. "Consistency Maintenance using UNIFY." University of Virginia Dept. of Computer Science Tech Report (1995).

University of Virginia, Department of Computer Science
Published Date: