Designing for Low-Latency Reads with Duplicate Covering Indexes
Let’s start this blog by considering a use case scenario that requires active-active application deployments across multiple regions supported by a YugabyteDB database cluster spanning three regions. While a multi-region YugabyteDB cluster provides region-level resiliency, how can you also ensure low-latency reads in this topology? Easy.
YugabyteDB features geographically located duplicate covering indexes that optimize for low-latency reads by ensuring they occur in the same region where an application is deployed. This is especially useful in a multi-region application and database deployment.
The design approach described below guarantees applications consistently experience low-latency, consistent reads while avoiding stale data reads. However, this design will have implications for write latency due to cross-region traffic and the need to write multiple copies across multiple regions.
A duplicate covering index is a type of covering index where the schema of the index is the same as the table. A covering index is an index that includes all the columns required by a query, allowing you to perform index-only scans.
