Add benchmark requests without session and alternating clients

[bdraco]

This was a bit of a blind spot in our benchmarks

noticed when I was looking at resolver object churn in #10847

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 98.70%. Comparing base (1caddcf) to head (ecbeda6).
Report is 2 commits behind head on master.

✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##           master   #10848   +/-   ##
=======================================
  Coverage   98.70%   98.70%           
=======================================
  Files         125      125           
  Lines       37858    37892   +34     
  Branches     2091     2094    +3     
=======================================
+ Hits        37366    37400   +34     
  Misses        343      343           
  Partials      149      149           

Flag	Coverage Δ
CI-GHA	98.58% <100.00%> (+<0.01%)	⬆️
OS-Linux	98.27% <100.00%> (+<0.01%)	⬆️
OS-Windows	96.41% <100.00%> (+0.01%)	⬆️
OS-macOS	97.39% <100.00%> (+<0.01%)	⬆️
Py-3.10.11	97.29% <100.00%> (+<0.01%)	⬆️
Py-3.10.17	97.61% <100.00%> (-0.22%)	⬇️
Py-3.11.12	97.92% <100.00%> (-0.01%)	⬇️
Py-3.11.9	97.38% <100.00%> (+0.01%)	⬆️
Py-3.12.10	98.35% <100.00%> (-0.01%)	⬇️
Py-3.13.3	98.34% <100.00%> (+<0.01%)	⬆️
Py-3.9.13	97.16% <100.00%> (+<0.01%)	⬆️
Py-3.9.22	97.52% <100.00%> (-0.19%)	⬇️
Py-pypy7.3.16	93.50% <100.00%> (+8.93%)	⬆️
VM-macos	97.39% <100.00%> (+<0.01%)	⬆️
VM-ubuntu	98.27% <100.00%> (+<0.01%)	⬆️
VM-windows	96.41% <100.00%> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[codspeed-hq]

CodSpeed Performance Report

Merging #10848 will not alter performance

_{Comparing benchmark_get_no_session (ecbeda6) with master (1caddcf)}

Summary

✅ 58 untouched benchmarks
🆕 2 new benchmarks

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
🆕	test_one_hundred_simple_get_requests_alternating_clients[pyloop]	N/A	32.9 ms	N/A
🆕	test_one_hundred_simple_get_requests_no_session[pyloop]	N/A	83.6 ms	N/A

[bdraco]

As suspected

And its reading files in the event loop 🙈

[Dreamsorcerer]

At the same time, I don't think users should expect high performance from the simple API. We tell them to use a session in production code and even do it from the very first example in the docs..

[bdraco]

At the same time, I don't think users should expect high performance from the simple API. We tell them to use a session in production code and even do it from the very first example in the docs..

I agree. We shouldn't put too much effort into improving this case. I think only the multi-session case matters

[bdraco]

Yeah its the multisession case that actually matters.

Testing Home Assistant startup on a production instance:

116 setups used the shared session
47 created their own session
1 didn't use a session

48 DNSResolver objects in memory

[bdraco]

Checked a few more instances using Home Assistant’s profiler.dump_log_objects service, which uses gc to inspect objects in memory:

• 2nd instance: 10 DNSResolver objects
• 3rd instance: 247 DNSResolver objects

The number of DNSResolver instances varies significantly depending on the integrations in use, as many integration create their own session as the underlying libs are not aware of the shared session HA offers. This definitely seems worth looking into to reduce unnecessary duplication.