Can we improve how we work with prepared queries and how they are cached
benjamin...@gmail.com
José Valim
Currently prepared queries are using references, and store a lot of data in ETS. In SQL we take a different approach to identify queries, we have globally unique ids, which are generated at compile time. I could imagine a world where the only information needed to be stored would be the conn pid and id to signify that the query has been prepared in the specific connection or needed to be prepared.To be clear, I'm not asking anyone to do this work, I can do it myself. But I want to gauge the core team if this kind of improvement is of intrest to them.
--
You received this message because you are subscribed to the Google Groups "elixir-ecto" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elixir-ecto...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/elixir-ecto/73b679f1-49e3-411d-a9d5-43fcf71aec18n%40googlegroups.com.
benjamin...@gmail.com
benjamin...@gmail.com
José Valim
To view this discussion visit https://groups.google.com/d/msgid/elixir-ecto/84ab3ed7-f6d9-4b19-8fe0-50608939f0can%40googlegroups.com.
benjamin...@gmail.com
CPU Information: Apple M1 Max
Number of Available Cores: 10
Available memory: 64 GB
Elixir 1.20.0-dev
Erlang 28.1
JIT enabled: true
Benchmark suite executing with the following configuration:
warmup: 2 s
time: 1 s
memory time: 2 s
reduction time: 2 s
parallel: 50
inputs: 1..100_000
Measured function call overhead as: 0 ns
Benchmarking ecto with input 1..100_000 ...
Benchmarking sql with input 1..100_000 ...
Calculating statistics...
Formatting results...
##### With input 1..100_000 #####
Name ips average deviation median 99th %
ecto 3.63 K 275.80 μs ±93.24% 190.79 μs 1065.27 μs
Comparison:
sql 107.00 K
ecto 3.63 K - 29.51x slower +266.46 μs
Memory usage statistics:
Name average deviation median 99th %
ecto 7 KB ±0.00% 7 KB 7 KB
Comparison:
sql 1.02 KB
ecto 7 KB - 4.53x memory usage +5.45 KB
Reduction count statistics:
Name Reduction count
sql 124
**All measurements for reduction count were the same**
benjamin...@gmail.com
And serialized:
CPU Information: Apple M1 Max
Number of Available Cores: 10
Available memory: 64 GB
Elixir 1.20.0-dev
Erlang 28.1
JIT enabled: true
Benchmark suite executing with the following configuration:
warmup: 2 s
time: 5 s
memory time: 2 s
reduction time: 2 s
parallel: 1
inputs: 1..100_000
Estimated total run time: 22 s
Measured function call overhead as: 0 ns
Benchmarking ecto with input 1..100_000 ...
Benchmarking sql with input 1..100_000 ...
Calculating statistics...
Formatting results...
##### With input 1..100_000 #####
Name ips average deviation median 99th %
ecto 184.57 K 5418.06 ns ±148.97% 5208 ns 6292 ns
Comparison:
sql 2130.22 K
ecto 184.57 K - 11.54x slower +4948.62 ns
Memory usage statistics:
Name Memory usage
sql 1.02 KB
**All measurements for memory usage were the same**
Reduction count statistics:
Name Reduction count
sql 124
**All measurements for reduction count were the same**
José Valim
To view this discussion visit https://groups.google.com/d/msgid/elixir-ecto/3327cc3e-ba96-4484-86c9-e50c2fdc29e3n%40googlegroups.com.
benjamin...@gmail.com
import Ecto.Query
SQL.Pool.start_link(%{name: :mypool, protocol: :tcp, size: 10})
defmodule SQL.Repo do
use Ecto.Repo, otp_app: :sql, adapter: Ecto.Adapters.Postgres
use SQL, adapter: SQL.Adapters.Postgres, repo: __MODULE__
def sql() do
{idx, _} = SQL.Pool.checkout(~SQL[select 1], :mypool)
SQL.Pool.checkin(idx, :mypool)
end
def ecto() do
checkout(fn -> to_sql(:all, select(from("users"), 1)) end)
end
end
Application.put_env(:sql, :ecto_repos, [SQL.Repo])
Application.put_env(:sql, SQL.Repo, log: false, username: "postgres", password: "postgres", hostname: "localhost", database: "sql_test#{System.get_env("MIX_TEST_PARTITION")}", pool_size: 10)
SQL.Repo.__adapter__().storage_up(SQL.Repo.config())
SQL.Repo.start_link()
Benchee.run(
%{
"sql" => fn _ -> SQL.Repo.sql() end,
"ecto" => fn _ -> SQL.Repo.ecto() end,
},
parallel: 50, time: 1,
inputs: %{"1..100_000" => Enum.to_list(1..100_000)},
memory_time: 2,
reduction_time: 2,
unit_scaling: :smallest,
measure_function_call_overhead: true,
profile_after: :eprof
)
import Ecto.Query
SQL.Pool.start_link(%{name: :mypool, protocol: :tcp, size: 10})
defmodule SQL.Repo do
use Ecto.Repo, otp_app: :sql, adapter: Ecto.Adapters.Postgres
use SQL, adapter: SQL.Adapters.Postgres, repo: __MODULE__
def sql() do
{idx, _} = SQL.Pool.checkout(SQL.parse("with recursive temp (n, fact) as (select 0, 1 union all select n+1, (n+1)*fact from temp where n < 9)"), :mypool)
SQL.Pool.checkin(idx, :mypool)
end
def ecto() do
checkout(fn -> SQL.parse("with recursive temp (n, fact) as (select 0, 1 union all select n+1, (n+1)*fact from temp where n < 9)") end)
end
end
Application.put_env(:sql, :ecto_repos, [SQL.Repo])
Application.put_env(:sql, SQL.Repo, log: false, username: "postgres", password: "postgres", hostname: "localhost", database: "sql_test#{System.get_env("MIX_TEST_PARTITION")}", pool_size: 10)
SQL.Repo.__adapter__().storage_up(SQL.Repo.config())
SQL.Repo.start_link()
Benchee.run(
%{
"sql" => fn _ -> SQL.Repo.sql() end,
"ecto" => fn _ -> SQL.Repo.ecto() end,
},
parallel: 50, time: 1,
inputs: %{"1..100_000" => Enum.to_list(1..100_000)},
memory_time: 2,
reduction_time: 2,
unit_scaling: :smallest,
measure_function_call_overhead: true,
profile_after: :eprof
)
CPU Information: Apple M1 Max
Number of Available Cores: 10
Available memory: 64 GB
Elixir 1.20.0-dev
Erlang 28.1
JIT enabled: true
Benchmark suite executing with the following configuration:
warmup: 2 s
time: 1 s
memory time: 2 s
reduction time: 2 s
parallel: 50
inputs: 1..100_000
Estimated total run time: 14 s
Measured function call overhead as: 0 ns
Benchmarking ecto with input 1..100_000 ...
Benchmarking sql with input 1..100_000 ...
Calculating statistics...
Formatting results...
##### With input 1..100_000 #####
Name ips average deviation median 99th %
ecto 2.96 K 337.61 μs ±116.37% 244 μs 1180.75 μs
Comparison:
sql 33.24 K
ecto 2.96 K - 11.22x slower +307.52 μs
Memory usage statistics:
Name Memory usage
sql 13.76 KB
ecto 16.20 KB - 1.18x memory usage +2.44 KB
**All measurements for memory usage were the same**
José Valim
To view this discussion visit https://groups.google.com/d/msgid/elixir-ecto/697a86e5-3d63-4abc-a165-5d9269b168efn%40googlegroups.com.
benjamin...@gmail.com
benjamin...@gmail.com
benjamin...@gmail.com
Now we have linear scaling of the pool, I've included an Ecto benchmark as well.
➜ sql git:(main) ✗ mix sql.bench.pool
🧠 SQL.Pool Live Benchmark | Concurrency: 40 | Scheduler Online: 10 | Scheduler: 10 | Total Queries: 46077 | QPS: 3125.0 | Total dead: 0 | Error Rate: 0.0% | Retry Rate: 0.0%
Time left: 0 s
Pool Size: Initial: 20 | Current: 20 | Recommended: 16 | Active: 20 | Idle: 0 | Dead: 0
λ (pool): 3125.0 req/s: ▆▆▆▇▇▆▆▇▇▇▆▇▇▇▆▆▆▆▆▆▇▇▇▆▇▇▇▆▇▆▇▇▇▇▆▇▇▆▇▇▇▇▇▆█▇▇▇▆▇
W: 6.0 ms: ▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇█▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇
L: 20 connections (recommended: 16): ██████████████████████████████████████████████████
Error Rate: 0.0% ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
Retry Rate: 0.0% ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
── Per-schedule utilization ─────────────
1: ██████████ 10%
↳ 2: ██████████████████████████████████████████████████ 99% (alive)
↳ 1: ██████████████████████████████████████████████████ 98% (alive)
2: ██████████ 10%
↳ 4: ██████████████████████████████████████████████████ 99% (alive)
↳ 3: ██████████████████████████████████████████████████ 99% (alive)
3: ██████████ 10%
↳ 6: ██████████████████████████████████████████████████ 99% (alive)
↳ 5: ██████████████████████████████████████████████████ 97% (alive)
4: ██████████ 10%
↳ 8: ██████████████████████████████████████████████████ 99% (alive)
↳ 7: ██████████████████████████████████████████████████ 98% (alive)
5: ██████████ 10%
↳ 9: ██████████████████████████████████████████████████ 98% (alive)
↳ 10: ██████████████████████████████████████████████████ 98% (alive)
6: ██████████ 10%
↳ 12: ██████████████████████████████████████████████████ 99% (alive)
↳ 11: ██████████████████████████████████████████████████ 97% (alive)
7: ██████████ 10%
↳ 14: ██████████████████████████████████████████████████ 100% (alive)
↳ 13: ██████████████████████████████████████████████████ 98% (alive)
8: ██████████ 10%
↳ 16: ██████████████████████████████████████████████████ 98% (alive)
↳ 15: ██████████████████████████████████████████████████ 98% (alive)
9: ██████████ 10%
↳ 17: ██████████████████████████████████████████████████ 97% (alive)
↳ 18: ██████████████████████████████████████████████████ 97% (alive)
10: ██████████ 10%
↳ 20: ██████████████████████████████████████████████████ 99% (alive)
↳ 19: ██████████████████████████████████████████████████ 96% (alive)
if System.monotonic_time() < stop_time do
case SQL.Pool.checkout(%{id: 1}, :default) do
{idx, _socket} ->
:counters.add(counter, 1, 1)
Process.sleep(Enum.random(1..10))
SQL.Pool.checkin(idx, :default)
:none -> Process.sleep(5)
end
loop_checkout(stop_time, counter, pid)
end
end
➜ sql git:(main) ✗ mix run /benchmarks/ecto.exs
🧠 SQL.Pool Live Benchmark | Concurrency: 40 | Total Queries: 22230 | QPS: 1566.7
Pool Size: Initial: 20
Time left: 0 s
if System.monotonic_time() < stop_time do
try do
Repo.checkout(fn ->
:counters.add(counter, 1, 1)
Process.sleep(Enum.random(1..10))
end)
rescue
_ -> :error
Process.sleep(5)
end
loop_checkout(stop_time, counter)
end
end
benjamin...@gmail.com
use Ecto.Repo, otp_app: :sql, adapter: Ecto.Adapters.Postgres
Application.put_env(:sql, :ecto_repos, [SQL.Repo])
Application.put_env(:sql, SQL.Repo, log: false, username: "postgres", password: "postgres", hostname: "localhost", database: "sql_test#{System.get_env("MIX_TEST_PARTITION")}", pool_size: 10)
SQL.Repo.__adapter__().storage_up(SQL.Repo.config())
@moduledoc false
@pool_size 40
@lease_time_ms 5
@duration_ms 5_000
def run do
{:ok, pid} = SQL.Repo.start_link()
IO.puts("Starting scaling benchmark (pool size: #{@pool_size})")
for clients <- 1..50 do
counter = :counters.new(1, [:write_concurrency])
stop_time = System.monotonic_time(:millisecond) + @duration_ms
Enum.each(1..clients, fn _ ->
spawn(fn -> loop_checkout(stop_time, pid, counter) end)
end)
# Wait for duration
Process.sleep(@duration_ms + 50)
total_requests = :counters.get(counter, 1)
measured_qps = total_requests / (@duration_ms / 1000)
theoretical_qps = @pool_size / (@lease_time_ms / 1000)
IO.puts("""
Clients: #{clients}
Total requests: #{total_requests}, Measured QPS: #{Float.round(measured_qps, 1)}
Theoretical max QPS: #{theoretical_qps |> Float.round(1)}
""")
end
end
defp loop_checkout(stop_time, pid, counter) do
if System.monotonic_time(:millisecond) < stop_time do
try do
SQL.Repo.checkout(fn ->
:counters.add(counter, 1, 1)
:rand.seed(:exsplus, {123, 456, 789})
delay = :rand.uniform(10) # 1..10 ms
Process.sleep(delay)
end)
rescue
_ -> :error
loop_checkout(stop_time, pid, counter)
end
end
end
SQL.Pool.DeterministicBench.run()
@moduledoc false
@pool_size 40
@lease_time_ms 5
@duration_ms 5_000
def run do
{:ok, pid} = SQL.Pool.start_link(%{name: :default, size: @pool_size, protocol: :tcp})
IO.puts("Starting scaling benchmark (pool size: #{@pool_size})")
for clients <- 1..50 do
counter = :counters.new(1, [:write_concurrency])
stop_time = System.monotonic_time(:millisecond) + @duration_ms
Enum.each(1..clients, fn _ ->
spawn(fn -> loop_checkout(stop_time, pid, counter) end)
end)
# Wait for duration
Process.sleep(@duration_ms + 50)
total_requests = :counters.get(counter, 1)
measured_qps = total_requests / (@duration_ms / 1000)
theoretical_qps = @pool_size / (@lease_time_ms / 1000)
IO.puts("""
Clients: #{clients}
Total requests: #{total_requests}, Measured QPS: #{Float.round(measured_qps, 1)}
Theoretical max QPS: #{theoretical_qps |> Float.round(1)}
""")
end
end
defp loop_checkout(stop_time, pid, counter) do
if System.monotonic_time(:millisecond) < stop_time do
{idx, _socket} = SQL.Pool.checkout(%{id: 1}, :default)
:counters.add(counter, 1, 1)
:rand.seed(:exsplus, {123, 456, 789})
delay = :rand.uniform(10) # 1..10 ms
Process.sleep(delay)
SQL.Pool.checkin(idx, :default)
loop_checkout(stop_time, pid, counter)
end
end
end
SQL.Pool.DeterministicBench.run()
➜ sql git:(main) ✗ mix run benchmarks/ecto.exs
Starting scaling benchmark (pool size: 40)
Clients: 1
Total requests: 803, Measured QPS: 160.6
Theoretical max QPS: 8.0e3
Clients: 2
Total requests: 1629, Measured QPS: 325.8
Theoretical max QPS: 8.0e3
Clients: 3
Total requests: 2441, Measured QPS: 488.2
Theoretical max QPS: 8.0e3
Clients: 4
Total requests: 3218, Measured QPS: 643.6
Theoretical max QPS: 8.0e3
Clients: 5
Total requests: 3999, Measured QPS: 799.8
Theoretical max QPS: 8.0e3
Clients: 6
Total requests: 4913, Measured QPS: 982.6
Theoretical max QPS: 8.0e3
Clients: 7
Total requests: 5728, Measured QPS: 1145.6
Theoretical max QPS: 8.0e3
Clients: 8
Total requests: 6533, Measured QPS: 1306.6
Theoretical max QPS: 8.0e3
Clients: 9
Total requests: 7311, Measured QPS: 1462.2
Theoretical max QPS: 8.0e3
Clients: 10
Total requests: 8152, Measured QPS: 1630.4
Theoretical max QPS: 8.0e3
Clients: 11
Total requests: 8084, Measured QPS: 1616.8
Theoretical max QPS: 8.0e3
Clients: 12
Total requests: 8193, Measured QPS: 1638.6
Theoretical max QPS: 8.0e3
Clients: 13
Total requests: 8095, Measured QPS: 1619.0
Theoretical max QPS: 8.0e3
Clients: 14
Total requests: 8145, Measured QPS: 1629.0
Theoretical max QPS: 8.0e3
Clients: 15
Total requests: 8182, Measured QPS: 1636.4
Theoretical max QPS: 8.0e3
Clients: 16
Total requests: 7884, Measured QPS: 1576.8
Theoretical max QPS: 8.0e3
Clients: 17
Total requests: 7966, Measured QPS: 1593.2
Theoretical max QPS: 8.0e3
Clients: 18
Total requests: 7510, Measured QPS: 1502.0
Theoretical max QPS: 8.0e3
Clients: 19
Total requests: 8055, Measured QPS: 1611.0
Theoretical max QPS: 8.0e3
Clients: 20
Total requests: 8104, Measured QPS: 1620.8
Theoretical max QPS: 8.0e3
Clients: 21
Total requests: 8125, Measured QPS: 1625.0
Theoretical max QPS: 8.0e3
Clients: 22
Total requests: 8132, Measured QPS: 1626.4
Theoretical max QPS: 8.0e3
Clients: 23
Total requests: 8079, Measured QPS: 1615.8
Theoretical max QPS: 8.0e3
Clients: 24
Total requests: 8107, Measured QPS: 1621.4
Theoretical max QPS: 8.0e3
Clients: 25
Total requests: 8130, Measured QPS: 1626.0
Theoretical max QPS: 8.0e3
Clients: 26
Total requests: 7683, Measured QPS: 1536.6
Theoretical max QPS: 8.0e3
Clients: 27
Total requests: 7894, Measured QPS: 1578.8
Theoretical max QPS: 8.0e3
Clients: 28
Total requests: 8015, Measured QPS: 1603.0
Theoretical max QPS: 8.0e3
Clients: 29
Total requests: 8062, Measured QPS: 1612.4
Theoretical max QPS: 8.0e3
Clients: 30
Total requests: 8024, Measured QPS: 1604.8
Theoretical max QPS: 8.0e3
Clients: 31
Total requests: 7965, Measured QPS: 1593.0
Theoretical max QPS: 8.0e3
Clients: 32
Total requests: 7837, Measured QPS: 1567.4
Theoretical max QPS: 8.0e3
Clients: 33
Total requests: 8025, Measured QPS: 1605.0
Theoretical max QPS: 8.0e3
Clients: 34
Total requests: 8064, Measured QPS: 1612.8
Theoretical max QPS: 8.0e3
Clients: 35
Total requests: 8077, Measured QPS: 1615.4
Theoretical max QPS: 8.0e3
Clients: 36
Total requests: 8191, Measured QPS: 1638.2
Theoretical max QPS: 8.0e3
Clients: 37
Total requests: 8104, Measured QPS: 1620.8
Theoretical max QPS: 8.0e3
Clients: 38
Total requests: 8138, Measured QPS: 1627.6
Theoretical max QPS: 8.0e3
Clients: 39
Total requests: 8103, Measured QPS: 1620.6
Theoretical max QPS: 8.0e3
Clients: 40
Total requests: 8223, Measured QPS: 1644.6
Theoretical max QPS: 8.0e3
Clients: 41
Total requests: 8178, Measured QPS: 1635.6
Theoretical max QPS: 8.0e3
Clients: 42
Total requests: 8196, Measured QPS: 1639.2
Theoretical max QPS: 8.0e3
Clients: 43
Total requests: 8176, Measured QPS: 1635.2
Theoretical max QPS: 8.0e3
Clients: 44
Total requests: 8107, Measured QPS: 1621.4
Theoretical max QPS: 8.0e3
Clients: 45
Total requests: 8186, Measured QPS: 1637.2
Theoretical max QPS: 8.0e3
Clients: 46
Total requests: 8247, Measured QPS: 1649.4
Theoretical max QPS: 8.0e3
Clients: 47
Total requests: 7920, Measured QPS: 1584.0
Theoretical max QPS: 8.0e3
Clients: 48
Total requests: 8057, Measured QPS: 1611.4
Theoretical max QPS: 8.0e3
Clients: 49
Total requests: 8159, Measured QPS: 1631.8
Theoretical max QPS: 8.0e3
Clients: 50
Total requests: 8123, Measured QPS: 1624.6
Theoretical max QPS: 8.0e3
➜ sql git:(main) ✗ mix sql.bench.pool
Starting scaling benchmark (pool size: 40)
Clients: 1
Total requests: 824, Measured QPS: 164.8
Theoretical max QPS: 8.0e3
Clients: 2
Total requests: 1645, Measured QPS: 329.0
Theoretical max QPS: 8.0e3
Clients: 3
Total requests: 2439, Measured QPS: 487.8
Theoretical max QPS: 8.0e3
Clients: 4
Total requests: 3216, Measured QPS: 643.2
Theoretical max QPS: 8.0e3
Clients: 5
Total requests: 4092, Measured QPS: 818.4
Theoretical max QPS: 8.0e3
Clients: 6
Total requests: 4860, Measured QPS: 972.0
Theoretical max QPS: 8.0e3
Clients: 7
Total requests: 5691, Measured QPS: 1138.2
Theoretical max QPS: 8.0e3
Clients: 8
Total requests: 6480, Measured QPS: 1296.0
Theoretical max QPS: 8.0e3
Clients: 9
Total requests: 7392, Measured QPS: 1478.4
Theoretical max QPS: 8.0e3
Clients: 10
Total requests: 8150, Measured QPS: 1630.0
Theoretical max QPS: 8.0e3
Clients: 11
Total requests: 8960, Measured QPS: 1792.0
Theoretical max QPS: 8.0e3
Clients: 12
Total requests: 9816, Measured QPS: 1963.2
Theoretical max QPS: 8.0e3
Clients: 13
Total requests: 10645, Measured QPS: 2129.0
Theoretical max QPS: 8.0e3
Clients: 14
Total requests: 11074, Measured QPS: 2214.8
Theoretical max QPS: 8.0e3
Clients: 15
Total requests: 12024, Measured QPS: 2404.8
Theoretical max QPS: 8.0e3
Clients: 16
Total requests: 12816, Measured QPS: 2563.2
Theoretical max QPS: 8.0e3
Clients: 17
Total requests: 13665, Measured QPS: 2733.0
Theoretical max QPS: 8.0e3
Clients: 18
Total requests: 14515, Measured QPS: 2903.0
Theoretical max QPS: 8.0e3
Clients: 19
Total requests: 15588, Measured QPS: 3117.6
Theoretical max QPS: 8.0e3
Clients: 20
Total requests: 16295, Measured QPS: 3259.0
Theoretical max QPS: 8.0e3
Clients: 21
Total requests: 17256, Measured QPS: 3451.2
Theoretical max QPS: 8.0e3
Clients: 22
Total requests: 18348, Measured QPS: 3669.6
Theoretical max QPS: 8.0e3
Clients: 23
Total requests: 19182, Measured QPS: 3836.4
Theoretical max QPS: 8.0e3
Clients: 24
Total requests: 20016, Measured QPS: 4003.2
Theoretical max QPS: 8.0e3
Clients: 25
Total requests: 20850, Measured QPS: 4170.0
Theoretical max QPS: 8.0e3
Clients: 26
Total requests: 21684, Measured QPS: 4336.8
Theoretical max QPS: 8.0e3
Clients: 27
Total requests: 22518, Measured QPS: 4503.6
Theoretical max QPS: 8.0e3
Clients: 28
Total requests: 23352, Measured QPS: 4670.4
Theoretical max QPS: 8.0e3
Clients: 29
Total requests: 24186, Measured QPS: 4837.2
Theoretical max QPS: 8.0e3
Clients: 30
Total requests: 25020, Measured QPS: 5004.0
Theoretical max QPS: 8.0e3
Clients: 31
Total requests: 25854, Measured QPS: 5170.8
Theoretical max QPS: 8.0e3
Clients: 32
Total requests: 26688, Measured QPS: 5337.6
Theoretical max QPS: 8.0e3
Clients: 33
Total requests: 27522, Measured QPS: 5504.4
Theoretical max QPS: 8.0e3
Clients: 34
Total requests: 28356, Measured QPS: 5671.2
Theoretical max QPS: 8.0e3
Clients: 35
Total requests: 29190, Measured QPS: 5838.0
Theoretical max QPS: 8.0e3
Clients: 36
Total requests: 30024, Measured QPS: 6004.8
Theoretical max QPS: 8.0e3
Clients: 37
Total requests: 30747, Measured QPS: 6149.4
Theoretical max QPS: 8.0e3
Clients: 38
Total requests: 31692, Measured QPS: 6338.4
Theoretical max QPS: 8.0e3
Clients: 39
Total requests: 32526, Measured QPS: 6505.2
Theoretical max QPS: 8.0e3
Clients: 40
Total requests: 33360, Measured QPS: 6672.0
Theoretical max QPS: 8.0e3
Clients: 41
Total requests: 34194, Measured QPS: 6838.8
Theoretical max QPS: 8.0e3
Clients: 42
Total requests: 35028, Measured QPS: 7005.6
Theoretical max QPS: 8.0e3
Clients: 43
Total requests: 35862, Measured QPS: 7172.4
Theoretical max QPS: 8.0e3
Clients: 44
Total requests: 36696, Measured QPS: 7339.2
Theoretical max QPS: 8.0e3
Clients: 45
Total requests: 37530, Measured QPS: 7506.0
Theoretical max QPS: 8.0e3
Clients: 46
Total requests: 38360, Measured QPS: 7672.0
Theoretical max QPS: 8.0e3
Clients: 47
Total requests: 39197, Measured QPS: 7839.4
Theoretical max QPS: 8.0e3
Clients: 48
Total requests: 40030, Measured QPS: 8006.0
Theoretical max QPS: 8.0e3
Clients: 49
Total requests: 40862, Measured QPS: 8172.4
Theoretical max QPS: 8.0e3
Clients: 50
Total requests: 41694, Measured QPS: 8338.8
Theoretical max QPS: 8.0e3