Design And Implementation Of A Distributed Financial Infrastructure Maintenance System

Inconsistent, inaccurate, or incomplete reference data is the primary cause for trade failures in the financial service industry. Exceptions in the trade lifecycle caused by unreliable reference data lead to increased operational risk, lost revenue opportunities, financial liability and the need for expensive manual trade duplication and reconciliation processes. Since artificial system delays are often introduced to make allowances for poor reference data systems, these systems are notorious for limiting the forward progress of automation and STP (Straight-through Processing).The distributed reference data maintenance system is a replacement for the original reference data system built in ad-hoc fashion. The system is targeted at a single entry, globally accessible, high-performance and low maintenance cost infrastructure with the ability to broadcast data change notifications.This paper covered the design, implementation and deployment details of the distributed data maintenance system in an international financial firm. It started by an examination of the status of the data plant prior to the project, and reached a summary of requirements we defined, with the design and implementation details. The paper also established the concept of the delay-based cache coherence. It then designed a coherence algorithm that met the requirements of the delay-based cache coherence. The paper ends with an analysis on the data flow and evolution plans on the distributed data maintenance system.