Estimating Data Loss At Scale

Abstract

For companies that serve corporate customers, Customer Service Outage (CSO) is a catastrophic event that may lead to some loss of their customer data. After each CSO, it is important to have a timely and quantitative measurement of how much data was lost. However, it is impractical for human to do so due to the enormous amount of data. In this paper, we present a robust solution that can return numerical loss report within hours. It handles a variety of challenges that are associated with the data. Consequently, management team can gauge the severity of data loss right after each event and respond accordingly.