correct!
We just updated the graphs to show throughput numbers instead. Should be more clear now.
This is true that the latency of each operation won't change much as we increase concurrent accesses, but the amortized latency would (be expected to) reduce as the graphs show. When we did benchmarks, the Go benchmark…
correct!
We just updated the graphs to show throughput numbers instead. Should be more clear now.
This is true that the latency of each operation won't change much as we increase concurrent accesses, but the amortized latency would (be expected to) reduce as the graphs show. When we did benchmarks, the Go benchmark…