Replace ISystemClock with TimeProvider in Kestrel

[Tratcher]

Fixes #47472
Fixes #13628

This replaces the internal ISystemClock types with the new TimeProvider abstraction. I also included #13628 which switches most operations to timestamps instead of utc dates, which may clean up the code a bit and allow for other micro-optimizations by working more directly with longs than DateTimeOffset.

Fixes #37862 (quarantined test which hasn't failed in 30+ days)

[danmoseley]

will we eventually have only one TestTimeProvider? looks like 4 are copies, plus the SignalR one is possibly mergeable

edit: never mind, you answered that already 😄

[Tratcher]

/benchmarks

[pr-benchmarks]

Crank Pull Request Bot

/benchmark <benchmark[,...]> <profile[,...]> <component,[...]> <arguments>

Benchmarks:

• plaintext: TechEmpower Plaintext Scenario - ASP.NET Platform implementation
• json: TechEmpower JSON Scenario - ASP.NET Platform implementation
• fortunes: TechEmpower Fortunes Scenario - ASP.NET Platform implementation
• yarp: YARP - http-http with 10 bytes
• mvcjsoninput2k: Sends 2Kb Json Body to an MVC controller

Profiles:

• aspnet-perf-lin: INTEL/Linux 12 Cores
• aspnet-perf-win: INTEL/Windows 12 Cores
• aspnet-citrine-lin: INTEL/Linux 28 Cores
• aspnet-citrine-win: INTEL/Windows 28 Cores
• aspnet-citrine-ampere: Ampere/Linux 80 Cores
• aspnet-citrine-amd: AMD/Linux 48 Cores

Components:

• kestrel
• mvc

Arguments: any additional arguments to pass through to crank, e.g. --variable name=value

[Tratcher]

/benchmark plaintext aspnet-citrine-lin kestrel

[pr-benchmarks]

Benchmark started for plaintext on aspnet-citrine-lin with kestrel. Logs: link

[Tratcher]

/benchmark plaintext aspnet-citrine-lin kestrel

[pr-benchmarks]

Benchmark started for plaintext on aspnet-citrine-lin with kestrel. Logs: link

[Tratcher]

will we eventually have only one TestTimeProvider? looks like 4 are copies, plus the SignalR one is possibly mergeable

I got rid of two custom time providers in this PR, one in product code and one test. Once this review is done I can look at reconciling across AspNetCore.

Update I got rid of all duplicate test time providers in AspNetCore.

[Tratcher]

/benchmark plaintext aspnet-citrine-lin kestrel

[pr-benchmarks]

Benchmark started for plaintext on aspnet-citrine-lin with kestrel. Logs: link

[sebastienros]

/benchmark plaintext aspnet-citrine-lin kestrel --application.sdkVersion 8.0.100-preview.5.23260.3

[pr-benchmarks]

Benchmark started for plaintext on aspnet-citrine-lin with kestrel and arguments --application.sdkVersion 8.0.100-preview.5.23260.3. Logs: link

[sebastienros]

| application           | plaintext.base            | plaintext.pr              |        |
| --------------------- | ------------------------- | ------------------------- | ------ |
| CPU Usage (%)         |                        99 |                        99 |  0.00% |
| Cores usage (%)       |                     2,774 |                     2,777 | +0.11% |
| Working Set (MB)      |                       122 |                       121 | -0.82% |
| Private Memory (MB)   |                       655 |                       651 | -0.61% |
| Build Time (ms)       |                     4,778 |                     4,378 | -8.37% |
| Start Time (ms)       |                       210 |                       210 |  0.00% |
| Published Size (KB)   |                    96,262 |                    96,262 |  0.00% |
| Symbols Size (KB)     |                        53 |                        53 |  0.00% |
| .NET Core SDK Version | 8.0.100-preview.5.23260.3 | 8.0.100-preview.5.23260.3 |        |


| load                   | plaintext.base | plaintext.pr |         |
| ---------------------- | -------------- | ------------ | ------- |
| CPU Usage (%)          |             93 |           94 |  +1.08% |
| Cores usage (%)        |          2,613 |        2,642 |  +1.11% |
| Working Set (MB)       |             48 |           48 |   0.00% |
| Private Memory (MB)    |            370 |          370 |   0.00% |
| Start Time (ms)        |              0 |            0 |         |
| First Request (ms)     |            103 |           99 |  -3.88% |
| Requests/sec           |     11,071,054 |   11,138,157 |  +0.61% |
| Requests               |    167,173,536 |  168,100,896 |  +0.55% |
| Mean latency (ms)      |           1.28 |         1.22 |  -4.69% |
| Max latency (ms)       |          61.79 |        91.46 | +48.02% |
| Bad responses          |              0 |            0 |         |
| Socket errors          |              0 |            0 |         |
| Read throughput (MB/s) |       1,331.20 |     1,341.44 |  +0.77% |
| Latency 50th (ms)      |           0.74 |         0.73 |  -1.34% |
| Latency 75th (ms)      |           1.13 |         1.10 |  -2.65% |
| Latency 90th (ms)      |           1.92 |         1.81 |  -5.73% |
| Latency 99th (ms)      |          14.03 |        13.36 |  -4.78% |

[Tratcher]

/benchmark plaintext aspnet-citrine-lin kestrel --application.sdkVersion 8.0.100-preview.5.23260.3

[pr-benchmarks]

Benchmark started for plaintext on aspnet-citrine-lin with kestrel and arguments --application.sdkVersion 8.0.100-preview.5.23260.3. Logs: link

[Tratcher]

| application           | plaintext.base            | plaintext.pr              |        |
| --------------------- | ------------------------- | ------------------------- | ------ |
| CPU Usage (%)         |                       100 |                        99 | -1.00% |
| Cores usage (%)       |                     2,787 |                     2,773 | -0.50% |
| Working Set (MB)      |                       124 |                       120 | -3.23% |
| Private Memory (MB)   |                       656 |                       651 | -0.76% |
| Build Time (ms)       |                     4,749 |                     4,289 | -9.69% |
| Start Time (ms)       |                       208 |                       207 | -0.48% |
| Published Size (KB)   |                    96,262 |                    96,262 |  0.00% |
| Symbols Size (KB)     |                        53 |                        53 |  0.00% |
| .NET Core SDK Version | 8.0.100-preview.5.23260.3 | 8.0.100-preview.5.23260.3 |        |


| load                   | plaintext.base | plaintext.pr |         |
| ---------------------- | -------------- | ------------ | ------- |
| CPU Usage (%)          |             94 |           95 |  +1.06% |
| Cores usage (%)        |          2,627 |        2,664 |  +1.41% |
| Working Set (MB)       |             48 |           48 |   0.00% |
| Private Memory (MB)    |            370 |          370 |   0.00% |
| Start Time (ms)        |              0 |            0 |         |
| First Request (ms)     |             92 |           96 |  +4.35% |
| Requests/sec           |     11,113,351 |   11,232,468 |  +1.07% |
| Requests               |    167,519,747 |  169,255,920 |  +1.04% |
| Mean latency (ms)      |           1.62 |         1.21 | -25.31% |
| Max latency (ms)       |         199.10 |        67.80 | -65.95% |
| Bad responses          |              0 |            0 |         |
| Socket errors          |              0 |            0 |         |
| Read throughput (MB/s) |       1,331.20 |     1,351.68 |  +1.54% |
| Latency 50th (ms)      |           0.74 |         0.73 |  -2.16% |
| Latency 75th (ms)      |           1.12 |         1.09 |  -2.68% |
| Latency 90th (ms)      |           1.88 |         1.74 |  -7.45% |
| Latency 99th (ms)      |          16.75 |        13.51 | -19.34% |

The few metrics that looked concerning are unstable across runs. I think this is good as far as perf goes.

[Tratcher]

/benchmark plaintext aspnet-citrine-lin kestrel --application.sdkVersion 8.0.100-preview.5.23260.3

[pr-benchmarks]

Benchmark started for plaintext on aspnet-citrine-lin with kestrel and arguments --application.sdkVersion 8.0.100-preview.5.23260.3. Logs: link