DECEMBER 07, 2022

RAMP-TAO: Layering Atomic Transactions on Meta’s Online TAO Data Store

Meta’s graph store TAO, like many other distributed data stores, has traditionally prioritized availability, efficiency, and horizontal scalability over strong consistency or isolation guarantees to serve its large read-dominant workloads. As product developers built diverse applications on top of this system, we have seen an increasing desire for transactional semantics. In this talk, we first characterize developer desires for transactions that emerged over the years and describe the current failure-atomic (i.e., write) transactions offered by TAO. We highlight the need for atomic visibility in a read transaction API with a measurement study on potential anomalies that occur without stronger isolation for reads. We then present the RAMP-TAO protocol, a variation based on the Read Atomic Multi-Partition (RAMP) protocol, that ensures atomic visibility for a read-optimized workload and an eventually consistent system. We demonstrate that this protocol results in minimal overhead, and our design can be feasibly deployed in production.

To help personalize content, tailor and measure ads, and provide a safer experience, we use cookies. By clicking or navigating the site, you agree to allow our collection of information on and off Facebook through cookies. Learn more, including about available controls: Cookies Policy