Atomic transaction chains for reliably updating partitioned databases

    Abstract

    To scale to millions of page requests per day, many popular web services such as Facebook, eBay, and Amazon partition application data, storing each partition on a separate database instance running on its own machine. Since these user-facing web services have strict latency and availability requirements, it costs too much to run distributed commit… (More)

    Topics

    9 Figures and Tables

    Slides referencing similar topics