Shadow Latency Overhead

Problem

Compute shadow deployment latency overhead as (mean_shadow - mean_primary) / mean_primary. Return 0.0 when either list is empty or the primary mean is 0.

Hint

Shadow traffic does not affect users but still consumes serving capacity.

Test cases

case 1

in: [[10,20],[12,26]]

expected: 0.26666666666666666

case 2

in: [[5],[5]]

expected: 0

case 3

in: [[0,0],[1,1]]

expected: 0

case 4

in: [[],[]]

expected: 0

Starter codePython

def shadow_latency_overhead(primary_latencies, shadow_latencies):
    # Your code here
    pass

4 test cases · run your code against them in the editor