Load Options

This is a walkthrough of different load options available to control the rate in requests per second (RPS) that ghz attempts to make to the server. All examples are done using a simple unary gRPC call.

Constant RPS

ghz --insecure --async \
  --proto /protos/helloworld.proto \
  --call helloworld.Greeter/SayHello \
  -c 10 -n 10000 --rps 200 \
  -d '{"name":"{{.WorkerID}}"}' 0.0.0.0:50051 

Summary:
  Count:    10000
  Total:    50.05 s
  Slowest:  56.17 ms
  Fastest:  50.17 ms
  Average:  50.58 ms
  Requests/sec: 199.79

Response time histogram:
  50.167 [1]    |
  50.768 [8056] |∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎
  51.368 [1900] |∎∎∎∎∎∎∎∎∎
  51.968 [37]   |
  52.568 [1]    |
  53.169 [0]    |
  53.769 [0]    |
  54.369 [1]    |
  54.969 [1]    |
  55.569 [0]    |
  56.170 [3]    |

Latency distribution:
  10 % in 50.32 ms
  25 % in 50.40 ms
  50 % in 50.56 ms
  75 % in 50.72 ms
  90 % in 50.88 ms
  95 % in 51.00 ms
  99 % in 51.23 ms

Status code distribution:
  [OK]   10000 responses

This will perform a constant load RPS against the server. Graphed, it may look like this:

Step Up RPS

ghz --insecure --async --proto /protos/helloworld.proto \
  --call helloworld.Greeter/SayHello \
  -c 10 -n 10000 \
  --load-schedule=step --load-start=50 --load-end=150 --load-step=10 --load-step-duration=5s \
  -d '{"name":"{{.WorkerID}}"}' 0.0.0.0:50051 

Summary:
  Count:    10000
  Total:    85.05 s
  Slowest:  60.16 ms
  Fastest:  50.18 ms
  Average:  51.10 ms
  Requests/sec: 117.57

Response time histogram:
  50.181 [1]    |
  51.179 [5713] |∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎
  52.177 [3923] |∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎
  53.174 [311]  |∎∎
  54.172 [46]   |
  55.170 [0]    |
  56.168 [1]    |
  57.166 [1]    |
  58.164 [2]    |
  59.161 [0]    |
  60.159 [2]    |

Latency distribution:
  10 % in 50.48 ms
  25 % in 50.68 ms
  50 % in 51.07 ms
  75 % in 51.41 ms
  90 % in 51.68 ms
  95 % in 52.02 ms
  99 % in 52.77 ms

Status code distribution:
  [OK]   10000 responses

Performs step load starting at 50 RPS and increasing by 10 RPS every 5s until we reach 150 RPS at which point the load is sustained at constant RPS rate until we reach 10000 total requests. The RPS load is distributed among the 10 workers, all sharing 1 connection.

Step Down RPS

ghz --insecure --async --proto /protos/helloworld.proto \
  --call helloworld.Greeter/SayHello \
  -c 10 -n 10000 \
  --load-schedule=step --load-start=200 --load-step=-10 --load-step-duration=5s --load-max-duration=40s \
  -d '{"name":"{{.WorkerID}}"}' 0.0.0.0:50051 

Summary:
  Count:    10000
  Total:    68.38 s
  Slowest:  55.88 ms
  Fastest:  50.16 ms
  Average:  50.85 ms
  Requests/sec: 146.23

Response time histogram:
  50.159 [1]    |
  50.730 [4367] |∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎
  51.302 [4281] |∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎
  51.874 [1304] |∎∎∎∎∎∎∎∎∎∎∎∎
  52.446 [43]   |
  53.018 [0]    |
  53.590 [0]    |
  54.162 [2]    |
  54.734 [0]    |
  55.306 [0]    |
  55.877 [2]    |

Latency distribution:
  10 % in 50.45 ms
  25 % in 50.60 ms
  50 % in 50.77 ms
  75 % in 51.06 ms
  90 % in 51.38 ms
  95 % in 51.54 ms
  99 % in 51.79 ms

Status code distribution:
  [OK]   10000 responses

Performs step load starting at 200 RPS and decreasing by 10 RPS every 10s until 40s has elapsed at which point the load is sustained at that RPS rate until we reach 10000 total requests. The RPS load is distributed among the 10 workers, all sharing 1 connection.

Linear load

ghz --insecure --async --proto /protos/helloworld.proto \
  --call helloworld.Greeter/SayHello \
  -c 10 -n 10000 \
  --load-schedule=line --load-start=5 --load-step=5 \
  -d '{"name":"{{.WorkerID}}"}' 0.0.0.0:50051

Summary:
  Count:    10000
  Total:    62.80 s
  Slowest:  56.61 ms
  Fastest:  50.19 ms
  Average:  50.72 ms
  Requests/sec: 159.24

Response time histogram:
  50.189 [1]    |
  50.832 [7552] |∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎
  51.474 [1781] |∎∎∎∎∎∎∎∎∎
  52.117 [405]  |∎∎
  52.759 [139]  |∎
  53.402 [42]   |
  54.044 [28]   |
  54.687 [28]   |
  55.329 [10]   |
  55.972 [13]   |
  56.614 [1]    |

Latency distribution:
  10 % in 50.34 ms
  25 % in 50.42 ms
  50 % in 50.57 ms
  75 % in 50.82 ms
  90 % in 51.22 ms
  95 % in 51.68 ms
  99 % in 53.16 ms

Status code distribution:
  [OK]   10000 responses

Performs linear load starting at 5 RPS and increasing by 5 RPS every 1s until we reach 10000 total requests. The RPS load is distributed among the 10 workers, all sharing 1 connection.

Linear load decrease

ghz --insecure --async --proto /protos/helloworld.proto \
  --call helloworld.Greeter/SayHello \
  -c 10 -n 10000 \
  --load-schedule=line --load-start=200 --load-step=-2 --load-end=100 \
  -d '{"name":"{{.WorkerID}}"}' 0.0.0.0:50051 

Summary:
  Count:    10000
  Total:    74.55 s
  Slowest:  83.20 ms
  Fastest:  50.18 ms
  Average:  50.89 ms
  Requests/sec: 134.13

Response time histogram:
  50.183 [1]    |
  53.486 [9989] |∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎∎
  56.788 [5]    |
  60.090 [1]    |
  63.392 [3]    |
  66.694 [0]    |
  69.996 [0]    |
  73.298 [0]    |
  76.600 [0]    |
  79.902 [0]    |
  83.205 [1]    |

Latency distribution:
  10 % in 50.42 ms
  25 % in 50.57 ms
  50 % in 50.77 ms
  75 % in 51.11 ms
  90 % in 51.49 ms
  95 % in 51.73 ms
  99 % in 52.15 ms

Status code distribution:
  [OK]   10000 responses

Performs linear load starting at 200 RPS and decreasing by 2 RPS every 1s until we reach 100 RPS at which point a constant rate is sustained until we accumulate 10000 total requests. The RPS load is distributed among the 10 workers, all sharing 1 connection.