Netflix Inc.

Vinay Chella

Joseph Lynch

Ajay Upadhyay

Ioannis Papapanagiotou

Various embodiments of the invention disclosed herein provide techniques for performing distributed anti-entropy repair procedures across a plurality of nodes in a distributed database network. A node included in a plurality of nodes within the distributed database network determines, before all other nodes included in the plurality of nodes, that a first anti-entropy repair procedure has ended. The node determines that a second anti-entropy repair procedure is ready to begin. The node generates a schedule for executing one or more operations associated with the second anti-entropy repair procedure. The node writes the schedule to a shared repair schedule data structure to initiate the second anti-entropy repair procedure across multiple nodes included in the plurality of nodes. Each of the nodes included in the plurality of nodes then performs a node repair based on the schedule.

Techniques for scheduled anti-entropy repair design

In various embodiments, a benchmarking engine automatically tests a data store to assess functionality and/or performance of the data store. The benchmarking engine generates data store operations based on dynamically adjustable configuration data. As the benchmarking engine generates the data store operations, the data store operations execute on the data store. In a complementary fashion, as the data store operations execute on the data store, the benchmarking engine generates statistics based on the results of the executed data store operations. Advantageously, because the benchmarking engine adjusts the number and/or type of data store operations that the benchmarking engine generates based on any changes to the configuration data, the workload that executes on the data store may be fine-tuned as the benchmarking engine executes.

Growing community of inventors

Santa Clara, CA, United States of America

Vinay Chella