[WIP][OTEL-1720] Add a new API to Concentrator that directly computes APM stats for OTLP traces

[songy23]

trafficstars

What does this PR do?

Motivation

Describe how to test/QA your changes

[pr-commenter[bot]]

Regression Detector

Regression Detector Results

Run ID: 3725f509-a34e-4acf-8c80-9556408c9149 Baseline: be8177a0766a87aca26ab8e319c70db1506f6525 Comparison: 3d4186b286cecc464e61e581bb8105b16096c913

Performance changes are noted in the perf column of each table:

• ✅ = significantly better comparison variant performance
• ❌ = significantly worse comparison variant performance
• ➖ = no significant change in performance

No significant changes in experiment optimization goals

Confidence level: 90.00% Effect size tolerance: |Δ mean %| ≥ 5.00%

There were no significant changes in experiment optimization goals at this confidence level and effect size tolerance.

Fine details of change detection per experiment

perf	experiment	goal	Δ mean %	Δ mean % CI
➖	tcp_syslog_to_blackhole	ingress throughput	+5.09	[-16.61, +26.78]
➖	idle	memory utilization	+0.55	[+0.50, +0.59]
➖	uds_dogstatsd_to_api_cpu	% cpu utilization	+0.23	[-2.71, +3.18]
➖	trace_agent_msgpack	ingress throughput	+0.01	[-0.01, +0.02]
➖	trace_agent_json	ingress throughput	+0.00	[-0.01, +0.01]
➖	otel_to_otel_logs	ingress throughput	+0.00	[-0.37, +0.37]
➖	uds_dogstatsd_to_api	ingress throughput	-0.01	[-0.21, +0.19]
➖	tcp_dd_logs_filter_exclude	ingress throughput	-0.03	[-0.06, +0.01]
➖	file_tree	memory utilization	-1.26	[-1.37, -1.15]
➖	pycheck_1000_100byte_tags	% cpu utilization	-1.34	[-5.87, +3.19]
➖	basic_py_check	% cpu utilization	-2.06	[-4.87, +0.75]

Explanation

A regression test is an A/B test of target performance in a repeatable rig, where "performance" is measured as "comparison variant minus baseline variant" for an optimization goal (e.g., ingress throughput). Due to intrinsic variability in measuring that goal, we can only estimate its mean value for each experiment; we report uncertainty in that value as a 90.00% confidence interval denoted "Δ mean % CI".

For each experiment, we decide whether a change in performance is a "regression" -- a change worth investigating further -- if all of the following criteria are true:

1. Its estimated |Δ mean %| ≥ 5.00%, indicating the change is big enough to merit a closer look.

2. Its 90.00% confidence interval "Δ mean % CI" does not contain zero, indicating that if our statistical model is accurate, there is at least a 90.00% chance there is a difference in performance between baseline and comparison variants.

3. Its configuration does not mark it "erratic".

[pr-commenter[bot]]

Test changes on VM

Use this command from test-infra-definitions to manually test this PR changes on a VM:

inv create-vm --pipeline-id=35227988 --os-family=ubuntu

[songy23]

Chatted offline, I'm going to try split this PR into incremental ones as it is currently too large.

[pkalmakis]

This feels like a very risky change as it refactors much of the existing code path for computing stats for all Datadog APM customers. Furthermore, it introduces a second forked code path for OTEL spans. Changing logic in the future will be much more difficult with these divergent code paths. Can we supply real-world evidence that the purported "overhead" of converting structs from one type to another justifies the an entirely separate code path? Furthermore, where can we find a test plan for ensuring this change doesn't create a regression for existing APM customers?

[songy23]

Chatted with @dineshg13 and @jdgumz offline and we agree to split out the OTel -> DD conversion code to a helper function, and use the same (DD) ingest API in Concentrator so that the changes in pkg/trace/stats are minimized.

This feels like a very risky change as it refactors much of the existing code path for computing stats for all Datadog APM customers. Furthermore, it introduces a second forked code path for OTEL spans.

@pkalmakis this is no longer the case in the revised approach. Now the new API only converts OTLP spans to Concentrator.Input rather than making any changes to Concentrator itself. APM stats processing on OTel is now identical to DD spans.

[pr-commenter[bot]]

Benchmarks

Benchmark execution time: 2024-05-22 22:05:42

Comparing candidate commit 80d28b2ac6d73739ebb27c667b7e0c32cd519f80 in PR branch yang.song/otlp-apm-stats with baseline commit 71ab04ee4581b7cf394ca85f8b8f7aa026482f9f in branch main.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 2 metrics, 1 unstable metrics.

[pr-commenter[bot]]

Benchmarks

Benchmark execution time: 2024-05-23 15:51:52

Comparing candidate commit f66446c2fc346f377ea2718703d85e1a7eab80d4 in PR branch yang.song/otlp-apm-stats with baseline commit 6520c3c1585049f70a248745107fdee728e7ae71 in branch main.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 2 metrics, 1 unstable metrics.

[pr-commenter[bot]]

Benchmarks

Benchmark execution time: 2024-05-23 15:54:59

Comparing candidate commit f66446c2fc346f377ea2718703d85e1a7eab80d4 in PR branch yang.song/otlp-apm-stats with baseline commit 6520c3c1585049f70a248745107fdee728e7ae71 in branch main.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 1 metrics, 2 unstable metrics.

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 91.51786% with 19 lines in your changes are missing coverage. Please review.

Project coverage is 51.98%. Comparing base (be8177a) to head (3d4186b). Report is 27 commits behind head on main.

Files	Patch %	Lines
pkg/trace/traceutil/otel_util.go	64.28%	10 Missing :warning:
pkg/trace/stats/otel_util.go	89.65%	6 Missing and 3 partials :warning:

Additional details and impacted files

@@             Coverage Diff             @@
##             main   #23503       +/-   ##
===========================================
+ Coverage   45.85%   51.98%    +6.13%     
===========================================
  Files        2181     1799      -382     
  Lines      256794   154027   -102767     
===========================================
- Hits       117747    80076    -37671     
+ Misses     129678    69485    -60193     
+ Partials     9369     4466     -4903     

Flag	Coverage Δ
amzn_aarch64	52.32% <91.51%> (+6.50%)	:arrow_up:
centos_x86_64	52.21% <91.51%> (+6.48%)	:arrow_up:
ubuntu_aarch64	52.33% <91.51%> (+6.50%)	:arrow_up:
ubuntu_x86_64	52.31% <91.51%> (+6.48%)	:arrow_up:
windows_amd64	56.03% <91.51%> (?)

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

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[pr-commenter[bot]]

Benchmarks

Benchmark execution time: 2024-05-23 20:11:47

Comparing candidate commit 53ddcfb20ab33ed8e2d41ddb743be314abbfdb6d in PR branch yang.song/otlp-apm-stats with baseline commit fa5ba1bb4c8fcdcee74136224e19d4ea55722e37 in branch main.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 2 metrics, 1 unstable metrics.

[pr-commenter[bot]]

Benchmarks

Benchmark execution time: 2024-05-23 21:30:29

Comparing candidate commit 289790ed1314304ef18d2a42926f6aa960bd4f19 in PR branch yang.song/otlp-apm-stats with baseline commit ece035dccfbae4701ca83b3ebe42807825005edd in branch main.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 2 metrics, 1 unstable metrics.

[github-actions[bot]]

Serverless Benchmark Results

BenchmarkStartEndInvocation comparison between 33d43600c674a65c10336b0edadc2b677c6604a0 and a5bca057782ea6fe14e0ab496843b617e4bdf2e3.

tl;dr

Use these benchmarks as an insight tool during development.

1. Skim down the vs base column in each chart. If there is a ~, then there was no statistically significant change to the benchmark. Otherwise, ensure the estimated percent change is either negative or very small.

2. The last row of each chart is the geomean. Ensure this percentage is either negative or very small.

What is this benchmarking?

The BenchmarkStartEndInvocation compares the amount of time it takes to call the start-invocation and end-invocation endpoints. For universal instrumentation languages (Dotnet, Golang, Java, Ruby), this represents the majority of the duration overhead added by our tracing layer.

The benchmark is run using a large variety of lambda request payloads. In the charts below, there is one row for each event payload type.

How do I interpret these charts?

The charts below comes from benchstat. They represent the statistical change in duration (sec/op), memory overhead (B/op), and allocations (allocs/op).

The benchstat docs explain how to interpret these charts.

Before the comparison table, we see common file-level configuration. If there are benchmarks with different configuration (for example, from different packages), benchstat will print separate tables for each configuration.

The table then compares the two input files for each benchmark. It shows the median and 95% confidence interval summaries for each benchmark before and after the change, and an A/B comparison under "vs base". ... The p-value measures how likely it is that any differences were due to random chance (i.e., noise). The "~" means benchstat did not detect a statistically significant difference between the two inputs. ...

Note that "statistically significant" is not the same as "large": with enough low-noise data, even very small changes can be distinguished from noise and considered statistically significant. It is, of course, generally easier to distinguish large changes from noise.

Finally, the last row of the table shows the geometric mean of each column, giving an overall picture of how the benchmarks changed. Proportional changes in the geomean reflect proportional changes in the benchmarks. For example, given n benchmarks, if sec/op for one of them increases by a factor of 2, then the sec/op geomean will increase by a factor of ⁿ√2.

I need more help

First off, do not worry if the benchmarks are failing. They are not tests. The intention is for them to be a tool for you to use during development.

If you would like a hand interpreting the results come chat with us in #serverless-agent in the internal DataDog slack or in #serverless in the public DataDog slack. We're happy to help!

Benchmark stats

goos: linux
goarch: amd64
pkg: github.com/DataDog/datadog-agent/pkg/serverless/daemon
cpu: AMD EPYC 7763 64-Core Processor                
                                      │ baseline/benchmark.log │        current/benchmark.log        │
                                      │         sec/op         │    sec/op     vs base               │
api-gateway-appsec.json                            85.90µ ± 8%    85.65µ ± 3%       ~ (p=0.971 n=10)
api-gateway-kong-appsec.json                       68.40µ ± 2%    66.17µ ± 1%  -3.25% (p=0.002 n=10)
api-gateway-kong.json                              67.19µ ± 1%    64.60µ ± 2%  -3.86% (p=0.000 n=10)
api-gateway-non-proxy-async.json                   105.8µ ± 1%    105.2µ ± 3%       ~ (p=0.436 n=10)
api-gateway-non-proxy.json                         104.1µ ± 2%    103.7µ ± 2%       ~ (p=0.912 n=10)
api-gateway-websocket-connect.json                 70.32µ ± 1%    69.67µ ± 2%       ~ (p=0.143 n=10)
api-gateway-websocket-default.json                 63.36µ ± 1%    61.93µ ± 2%  -2.25% (p=0.023 n=10)
api-gateway-websocket-disconnect.json              63.26µ ± 1%    61.95µ ± 2%  -2.08% (p=0.023 n=10)
api-gateway.json                                   115.7µ ± 1%    113.4µ ± 1%  -1.99% (p=0.003 n=10)
application-load-balancer.json                     61.94µ ± 1%    62.50µ ± 1%  +0.90% (p=0.019 n=10)
cloudfront.json                                    47.08µ ± 2%    47.90µ ± 3%  +1.75% (p=0.011 n=10)
cloudwatch-events.json                             37.31µ ± 2%    38.48µ ± 3%  +3.14% (p=0.002 n=10)
cloudwatch-logs.json                               64.80µ ± 1%    66.41µ ± 2%  +2.48% (p=0.002 n=10)
custom.json                                        29.25µ ± 5%    30.84µ ± 2%  +5.44% (p=0.002 n=10)
dynamodb.json                                      92.24µ ± 2%    93.76µ ± 1%  +1.65% (p=0.019 n=10)
empty.json                                         27.71µ ± 3%    29.18µ ± 2%  +5.30% (p=0.001 n=10)
eventbridge-custom.json                            40.59µ ± 2%    42.37µ ± 3%  +4.37% (p=0.000 n=10)
http-api.json                                      71.39µ ± 3%    74.39µ ± 2%  +4.21% (p=0.000 n=10)
kinesis-batch.json                                 70.42µ ± 2%    71.86µ ± 2%  +2.04% (p=0.009 n=10)
kinesis.json                                       52.70µ ± 2%    55.00µ ± 2%  +4.37% (p=0.000 n=10)
s3.json                                            58.37µ ± 2%    59.75µ ± 1%  +2.38% (p=0.001 n=10)
sns-batch.json                                     89.12µ ± 2%    91.14µ ± 2%  +2.27% (p=0.043 n=10)
sns.json                                           62.92µ ± 1%    64.07µ ± 3%  +1.82% (p=0.011 n=10)
snssqs.json                                        109.9µ ± 2%    111.9µ ± 2%       ~ (p=0.075 n=10)
snssqs_no_dd_context.json                          97.69µ ± 1%   100.28µ ± 2%  +2.66% (p=0.000 n=10)
sqs-aws-header.json                                53.82µ ± 1%    55.84µ ± 3%  +3.75% (p=0.000 n=10)
sqs-batch.json                                     93.45µ ± 2%    96.55µ ± 2%  +3.32% (p=0.004 n=10)
sqs.json                                           67.10µ ± 2%    69.68µ ± 2%  +3.84% (p=0.000 n=10)
sqs_no_dd_context.json                             60.68µ ± 2%    62.45µ ± 2%  +2.91% (p=0.000 n=10)
geomean                                            65.98µ         66.97µ       +1.51%

                                      │ baseline/benchmark.log │        current/benchmark.log        │
                                      │          B/op          │     B/op      vs base               │
api-gateway-appsec.json                           37.13Ki ± 0%   37.13Ki ± 0%       ~ (p=0.697 n=10)
api-gateway-kong-appsec.json                      26.80Ki ± 0%   26.80Ki ± 0%       ~ (p=0.839 n=10)
api-gateway-kong.json                             24.29Ki ± 0%   24.28Ki ± 0%       ~ (p=0.670 n=10)
api-gateway-non-proxy-async.json                  47.93Ki ± 0%   47.94Ki ± 0%       ~ (p=0.698 n=10)
api-gateway-non-proxy.json                        47.15Ki ± 0%   47.14Ki ± 0%       ~ (p=0.987 n=10)
api-gateway-websocket-connect.json                25.37Ki ± 0%   25.37Ki ± 0%       ~ (p=0.617 n=10)
api-gateway-websocket-default.json                21.27Ki ± 0%   21.27Ki ± 0%       ~ (p=0.362 n=10)
api-gateway-websocket-disconnect.json             21.05Ki ± 0%   21.05Ki ± 0%       ~ (p=0.697 n=10)
api-gateway.json                                  49.45Ki ± 0%   49.45Ki ± 0%       ~ (p=0.240 n=10)
application-load-balancer.json                    22.23Ki ± 0%   22.23Ki ± 0%       ~ (p=0.591 n=10)
cloudfront.json                                   17.58Ki ± 0%   17.57Ki ± 0%       ~ (p=0.382 n=10)
cloudwatch-events.json                            11.61Ki ± 0%   11.62Ki ± 0%  +0.09% (p=0.045 n=10)
cloudwatch-logs.json                              53.28Ki ± 0%   53.30Ki ± 0%       ~ (p=0.288 n=10)
custom.json                                       9.640Ki ± 0%   9.652Ki ± 0%  +0.13% (p=0.017 n=10)
dynamodb.json                                     40.59Ki ± 0%   40.57Ki ± 0%       ~ (p=0.060 n=10)
empty.json                                        9.209Ki ± 0%   9.208Ki ± 0%       ~ (p=0.470 n=10)
eventbridge-custom.json                           13.32Ki ± 0%   13.33Ki ± 0%       ~ (p=0.089 n=10)
http-api.json                                     23.62Ki ± 0%   23.65Ki ± 0%  +0.15% (p=0.000 n=10)
kinesis-batch.json                                26.94Ki ± 0%   26.95Ki ± 0%       ~ (p=0.255 n=10)
kinesis.json                                      17.71Ki ± 0%   17.75Ki ± 0%  +0.20% (p=0.006 n=10)
s3.json                                           20.22Ki ± 0%   20.26Ki ± 0%       ~ (p=0.052 n=10)
sns-batch.json                                    38.55Ki ± 0%   38.56Ki ± 0%       ~ (p=0.494 n=10)
sns.json                                          23.88Ki ± 0%   23.90Ki ± 0%       ~ (p=0.724 n=10)
snssqs.json                                       50.54Ki ± 0%   50.53Ki ± 0%       ~ (p=0.469 n=10)
snssqs_no_dd_context.json                         44.71Ki ± 0%   44.74Ki ± 0%  +0.07% (p=0.006 n=10)
sqs-aws-header.json                               18.77Ki ± 0%   18.79Ki ± 0%       ~ (p=0.699 n=10)
sqs-batch.json                                    41.56Ki ± 0%   41.60Ki ± 0%       ~ (p=0.305 n=10)
sqs.json                                          25.42Ki ± 0%   25.43Ki ± 0%       ~ (p=0.644 n=10)
sqs_no_dd_context.json                            20.67Ki ± 1%   20.60Ki ± 0%       ~ (p=0.436 n=10)
geomean                                           25.60Ki        25.61Ki       +0.03%

                                      │ baseline/benchmark.log │        current/benchmark.log        │
                                      │       allocs/op        │ allocs/op   vs base                 │
api-gateway-appsec.json                             629.0 ± 0%   629.5 ± 0%       ~ (p=1.000 n=10)
api-gateway-kong-appsec.json                        488.0 ± 0%   488.0 ± 0%       ~ (p=1.000 n=10) ¹
api-gateway-kong.json                               466.0 ± 0%   466.0 ± 0%       ~ (p=1.000 n=10) ¹
api-gateway-non-proxy-async.json                    725.0 ± 0%   725.0 ± 0%       ~ (p=0.628 n=10)
api-gateway-non-proxy.json                          716.0 ± 0%   716.0 ± 0%       ~ (p=1.000 n=10)
api-gateway-websocket-connect.json                  453.0 ± 0%   453.0 ± 0%       ~ (p=1.000 n=10) ¹
api-gateway-websocket-default.json                  379.0 ± 0%   379.0 ± 0%       ~ (p=1.000 n=10) ¹
api-gateway-websocket-disconnect.json               369.0 ± 0%   369.0 ± 0%       ~ (p=0.628 n=10)
api-gateway.json                                    790.0 ± 0%   790.0 ± 0%       ~ (p=0.474 n=10)
application-load-balancer.json                      352.0 ± 0%   351.0 ± 0%       ~ (p=0.656 n=10)
cloudfront.json                                     284.0 ± 0%   283.0 ± 0%       ~ (p=0.656 n=10)
cloudwatch-events.json                              220.0 ± 0%   220.0 ± 0%       ~ (p=1.000 n=10) ¹
cloudwatch-logs.json                                215.0 ± 0%   215.0 ± 0%       ~ (p=1.000 n=10)
custom.json                                         168.0 ± 0%   168.0 ± 0%       ~ (p=0.526 n=10)
dynamodb.json                                       589.0 ± 0%   588.0 ± 0%       ~ (p=0.370 n=10)
empty.json                                          159.0 ± 1%   159.0 ± 1%       ~ (p=0.628 n=10)
eventbridge-custom.json                             253.0 ± 0%   254.0 ± 0%       ~ (p=0.179 n=10)
http-api.json                                       432.0 ± 0%   432.5 ± 0%       ~ (p=0.094 n=10)
kinesis-batch.json                                  390.0 ± 0%   391.0 ± 0%       ~ (p=0.370 n=10)
kinesis.json                                        284.5 ± 0%   285.0 ± 0%  +0.18% (p=0.022 n=10)
s3.json                                             357.0 ± 0%   358.0 ± 0%  +0.28% (p=0.050 n=10)
sns-batch.json                                      454.0 ± 0%   454.5 ± 0%       ~ (p=0.650 n=10)
sns.json                                            322.0 ± 0%   323.0 ± 0%       ~ (p=0.611 n=10)
snssqs.json                                         446.0 ± 0%   446.0 ± 0%       ~ (p=1.000 n=10)
snssqs_no_dd_context.json                           399.0 ± 0%   399.0 ± 0%   0.00% (p=0.010 n=10)
sqs-aws-header.json                                 273.0 ± 0%   273.5 ± 1%       ~ (p=0.651 n=10)
sqs-batch.json                                      503.0 ± 0%   504.0 ± 0%       ~ (p=0.259 n=10)
sqs.json                                            350.0 ± 0%   350.0 ± 0%       ~ (p=0.834 n=10)
sqs_no_dd_context.json                              325.5 ± 1%   324.5 ± 0%       ~ (p=0.421 n=10)
geomean                                             376.2        376.3       +0.03%
¹ all samples are equal

[pr-commenter[bot]]

Benchmarks

Benchmark execution time: 2024-05-24 00:44:45

Comparing candidate commit 4a4578f219a745acbe1aa556de3c142032eacf21 in PR branch yang.song/otlp-apm-stats with baseline commit 33d43600c674a65c10336b0edadc2b677c6604a0 in branch main.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 2 metrics, 1 unstable metrics.

[pr-commenter[bot]]

Benchmarks

Benchmark execution time: 2024-05-24 19:48:13

Comparing candidate commit ebf21694f40505d166251059bba1b2189bd0df3d in PR branch yang.song/otlp-apm-stats with baseline commit 7ac9c8b8595235bdf9a3ed457d0095100251c451 in branch main.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 2 metrics, 1 unstable metrics.

[songy23]

/merge

[dd-devflow[bot]]

:steam_locomotive: MergeQueue

This merge request is not mergeable yet, because of pending checks/missing approvals. It will be added to the queue as soon as checks pass and/or get approvals. Note: if you pushed new commits since the last approval, you may need additional approval. You can remove it from the waiting list with /remove command.

Use /merge -c to cancel this operation!

[pr-commenter[bot]]

Benchmarks

Benchmark execution time: 2024-05-24 22:11:53

Comparing candidate commit 13a6fe883701ece1b8a2048f2c4637d5c4ea9a90 in PR branch yang.song/otlp-apm-stats with baseline commit e19773f96565662a14fad551d5be0dee5e668ef7 in branch main.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 2 metrics, 1 unstable metrics.

[dd-devflow[bot]]

:warning: MergeQueue

This merge request was unqueued

If you need support, contact us on Slack #devflow!

[songy23]

/merge

[dd-devflow[bot]]

:steam_locomotive: MergeQueue

This merge request is not mergeable yet, because of pending checks/missing approvals. It will be added to the queue as soon as checks pass and/or get approvals. Note: if you pushed new commits since the last approval, you may need additional approval. You can remove it from the waiting list with /remove command.

Use /merge -c to cancel this operation!

[pr-commenter[bot]]

Benchmarks

Benchmark execution time: 2024-05-28 12:54:40

Comparing candidate commit 3d4186b286cecc464e61e581bb8105b16096c913 in PR branch yang.song/otlp-apm-stats with baseline commit be8177a0766a87aca26ab8e319c70db1506f6525 in branch main.

Found 0 performance improvements and 0 performance regressions! Performance is the same for 2 metrics, 1 unstable metrics.

[dd-devflow[bot]]

:warning: MergeQueue

This merge request was unqueued

If you need support, contact us on Slack #devflow!

[songy23]

/merge

[dd-devflow[bot]]

:steam_locomotive: MergeQueue

This merge request is not mergeable yet, because of pending checks/missing approvals. It will be added to the queue as soon as checks pass and/or get approvals. Note: if you pushed new commits since the last approval, you may need additional approval. You can remove it from the waiting list with /remove command.

Use /merge -c to cancel this operation!

[songy23]

/merge -c

[dd-devflow[bot]]

:warning: MergeQueue

This merge request was unqueued

If you need support, contact us on Slack #devflow!

[songy23]

/merge

[dd-devflow[bot]]

:steam_locomotive: MergeQueue

Pull request added to the queue.

There are 4 builds ahead! (estimated merge in less than 2h)

Use /merge -c to cancel this operation!