We aim to compare behavior across traditional (single-node) DataFusion and distributed setups with 1, 2, and 3 workers. Although local benchmarking doesn't yield highly accurate performance metrics for distributed systems, it provides useful insights into system behavior and helps surface early performance regressions.
With the availability of cluster setup for end-to-end test, this work might be not that that tricky
We aim to compare behavior across traditional (single-node) DataFusion and distributed setups with 1, 2, and 3 workers. Although local benchmarking doesn't yield highly accurate performance metrics for distributed systems, it provides useful insights into system behavior and helps surface early performance regressions.
With the availability of cluster setup for end-to-end test, this work might be not that that tricky