Developed a performance benchmarking suite for the zarr-python repository, focusing on store indexing under varied latency conditions. The work introduced a LatencyStore component to simulate realistic latency scenarios, enabling comprehensive stress-testing of both sharded and local store indexing paths. By parameterizing benchmarks for different store types and latency values, the solution provided a robust framework for ongoing performance health checks and regression detection. Implemented entirely in Python, the approach emphasized performance benchmarking and testing best practices, laying the groundwork for improved reliability and visibility into indexing performance for users relying on zarr’s data storage and retrieval capabilities.