Fix benchmark issues

[jbertran]

What does this PR do?

Service naming (#2941) is the culprit for the recent benchmark issues, this PR tries to identify the cause and fix it.

[github-actions]

Overall package size

Self size: 4.13 MB
Deduped: 58.26 MB
No deduping: 58.31 MB

Dependency sizes

name	version	self size	total size
@datadog/pprof	2.2.1	14.24 MB	15.12 MB
@datadog/native-iast-taint-tracking	1.4.1	14.85 MB	14.86 MB
@datadog/native-appsec	3.1.0	13.31 MB	13.32 MB
protobufjs	7.1.2	2.76 MB	6.55 MB
@datadog/native-iast-rewriter	2.0.1	2.09 MB	2.1 MB
@datadog/native-metrics	2.0.0	898.77 kB	1.3 MB
opentracing	0.14.7	194.81 kB	194.81 kB
semver	7.3.8	88.2 kB	118.6 kB
@datadog/sketches-js	2.1.0	109.9 kB	109.9 kB
lodash.sortby	4.7.0	75.76 kB	75.76 kB
lru-cache	7.14.0	74.95 kB	74.95 kB
ipaddr.js	2.0.1	59.52 kB	59.52 kB
ignore	5.2.0	48.87 kB	48.87 kB
import-in-the-middle	1.3.5	34.34 kB	38.81 kB
istanbul-lib-coverage	3.2.0	29.34 kB	29.34 kB
retry	0.10.1	27.44 kB	27.44 kB
lodash.uniq	4.5.0	25.01 kB	25.01 kB
limiter	1.1.5	23.17 kB	23.17 kB
lodash.kebabcase	4.1.1	17.75 kB	17.75 kB
lodash.pick	4.4.0	16.33 kB	16.33 kB
node-abort-controller	3.0.1	14.33 kB	14.33 kB
crypto-randomuuid	1.0.0	11.18 kB	11.18 kB
diagnostics_channel	1.1.0	7.07 kB	7.07 kB
path-to-regexp	0.1.7	6.78 kB	6.78 kB
koalas	1.0.2	6.47 kB	6.47 kB
methods	1.1.2	5.29 kB	5.29 kB
module-details-from-path	1.0.3	4.47 kB	4.47 kB

_{🤖 This report was automatically generated by heaviest-objects-in-the-universe}

[codecov]

Codecov Report

Merging #3099 (db835c6) into master (cad3d92) will decrease coverage by 0.06%.
The diff coverage is 59.18%.

@@            Coverage Diff             @@
##           master    #3099      +/-   ##
==========================================
- Coverage   86.92%   86.86%   -0.06%     
==========================================
  Files         318      327       +9     
  Lines       11630    11810     +180     
  Branches       33       33              
==========================================
+ Hits        10109    10259     +150     
- Misses       1521     1551      +30     

Impacted Files	Coverage Δ
packages/dd-trace/src/plugins/outgoing.js	45.45% <0.00%> (-4.55%)	⬇️
packages/dd-trace/src/plugins/tracing.js	55.55% <16.66%> (-7.78%)	⬇️
.../dd-trace/src/service-naming/schemas/definition.js	28.57% <28.57%> (ø)
packages/dd-trace/src/service-naming/schemas/v0.js	42.85% <42.85%> (ø)
packages/dd-trace/src/service-naming/schemas/v1.js	50.00% <50.00%> (ø)
packages/dd-trace/src/service-naming/index.js	71.42% <71.42%> (ø)
packages/datadog-plugin-rhea/src/consumer.js	100.00% <100.00%> (ø)
packages/datadog-plugin-rhea/src/producer.js	100.00% <100.00%> (ø)
packages/dd-trace/src/config.js	98.98% <100.00%> (+0.04%)	⬆️
packages/dd-trace/src/plugin_manager.js	98.46% <100.00%> (+0.04%)	⬆️
... and 1 more

... and 4 files with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[pr-commenter]

Benchmarks

Comparing candidate commit db835c6 in PR branch jbertran/fix-svc-naming-perfs with baseline commit cad3d92 in branch master.

Found 0 performance improvements and 13 performance regressions! Performance is the same for 432 metrics, 27 unstable metrics.

scenario:log-without-log-16

• 🟥 execution_time [+9.859ms; +10.583ms] or [+5.743%; +6.164%]

scenario:log-with-error-16

• 🟥 execution_time [+9.924ms; +11.547ms] or [+5.721%; +6.657%]

scenario:log-skip-log-16

• 🟥 execution_time [+10.155ms; +11.099ms] or [+5.857%; +6.402%]

scenario:log-with-debug-16

• 🟥 execution_time [+8.740ms; +11.151ms] or [+5.013%; +6.396%]

scenario:startup-with-tracer-16

• 🟥 execution_time [+9.705ms; +10.537ms] or [+6.065%; +6.585%]

scenario:spans-finish-immediately-16

• 🟥 cpu_user_time [+29.138ms; +35.268ms] or [+7.492%; +9.069%]
• 🟥 execution_time [+33.356ms; +36.181ms] or [+7.981%; +8.657%]

scenario:log-without-log-18

• 🟥 execution_time [+9.007ms; +13.143ms] or [+5.100%; +7.442%]

scenario:log-with-debug-18

• 🟥 cpu_user_time [+7.538ms; +14.845ms] or [+5.034%; +9.913%]
• 🟥 execution_time [+10.291ms; +11.167ms] or [+5.810%; +6.304%]

scenario:log-with-error-18

• 🟥 execution_time [+9.918ms; +11.764ms] or [+5.589%; +6.629%]

scenario:log-skip-log-18

• 🟥 execution_time [+8.996ms; +11.102ms] or [+5.042%; +6.222%]

scenario:startup-with-tracer-18

• 🟥 execution_time [+9.794ms; +10.390ms] or [+5.912%; +6.272%]

[tlhunter]

Maybe try commenting out just the changes in plugin manager, test that, comment out just hte changes in config, test that, etc.

Tests will of course fail but see if the benchmark results improve.

[tlhunter]

Hopefully #3150 fixes the log benchmark issues. I haven't yet tried to fix the startup issues.

[tlhunter]

I'm rebasing on master now that some benchmark changes landed. Hopefully the log benchmark issues disappear.

[tlhunter]

Apparently #3150 had no affect?! The results are the same.

[jbertran]

Comparing with the previous benchmark, we've removed cpu_user_time ballooning for most of the log benchmarks. #3146, which I've also rebased, exhibits the same behaviour. I guess this would point towards the performance hit spending significant time in system calls, where a log spends most of its time in user calls?

The interesting part for the log benchmark is that the without-log scenario is showing the same performance hit as the rest, so it's possible that none of the scenarios actually log messages for some reason.

[jbertran]

We're reasonably certain the benchmarks aren't representative. This is no longer needed.