Fix audit findings for coverage, architecture checks, and XML docs
This commit is contained in:
@@ -1,6 +1,6 @@
|
||||
# CBDD Performance Benchmarks
|
||||
|
||||
> **Last Updated:** February 13, 2026
|
||||
> **Last Updated:** February 20, 2026
|
||||
> **Platform:** Windows 11, Intel Core i7-13800H (14 cores), .NET 10.0
|
||||
|
||||
---
|
||||
@@ -164,12 +164,59 @@ dotnet run -c Release --project tests/CBDD.Tests.Benchmark
|
||||
| 'Deserialize List 10k (JSON loop)' | 42,961.024 μs | 534.7024 μs | 446.5008 μs | 5333.3333 | 66944224 B |
|
||||
```
|
||||
|
||||
### Full Insert Benchmark Output
|
||||
|
||||
```
|
||||
| Method | Mean | Error | StdDev | Ratio | Allocated |
|
||||
|------------------------------------ |-----------:|----------:|----------:|------:|----------:|
|
||||
| 'SQLite Single Insert (AutoCommit)' | 2,916.3 μs | 130.50 μs | 382.73 μs | 1.00 | 6.67 KB |
|
||||
| 'DocumentDb Single Insert' | 355.8 μs | 19.42 μs | 56.65 μs | 0.12 | 128.89 KB |
|
||||
```
|
||||
### Full Insert Benchmark Output
|
||||
|
||||
```
|
||||
| Method | Mean | Error | StdDev | Ratio | Allocated |
|
||||
|------------------------------------ |-----------:|----------:|----------:|------:|----------:|
|
||||
| 'SQLite Single Insert (AutoCommit)' | 2,916.3 μs | 130.50 μs | 382.73 μs | 1.00 | 6.67 KB |
|
||||
| 'DocumentDb Single Insert' | 355.8 μs | 19.42 μs | 56.65 μs | 0.12 | 128.89 KB |
|
||||
```
|
||||
|
||||
---
|
||||
|
||||
## Performance Gate Evidence (compact_plan.md §11.4)
|
||||
|
||||
This section captures the required evidence for plan gate `11.4`:
|
||||
- documented write/read overhead
|
||||
- documented compaction throughput
|
||||
- no obvious pathological GC spikes for compression-enabled workloads
|
||||
|
||||
### Repro Commands
|
||||
|
||||
```bash
|
||||
# Benchmark smoke (write/read + serialization)
|
||||
dotnet run -c Release --project tests/CBDD.Tests.Benchmark/ZB.MOM.WW.CBDD.Tests.Benchmark.csproj
|
||||
|
||||
# Performance gate smoke (compaction throughput + compression GC deltas)
|
||||
dotnet run -c Release --project tests/CBDD.Tests.Benchmark/ZB.MOM.WW.CBDD.Tests.Benchmark.csproj -- gate
|
||||
```
|
||||
|
||||
### Captured Artifacts
|
||||
|
||||
- `BenchmarkDotNet.Artifacts/results/ZB.MOM.WW.CBDD.Tests.Benchmark.InsertBenchmarks-report.csv`
|
||||
- `BenchmarkDotNet.Artifacts/results/ZB.MOM.WW.CBDD.Tests.Benchmark.ReadBenchmarks-report.csv`
|
||||
- `BenchmarkDotNet.Artifacts/results/PerformanceGateSmoke-report.json`
|
||||
|
||||
### Captured Values (2026-02-20 UTC)
|
||||
|
||||
| Metric | Value | Source |
|
||||
|:--|--:|:--|
|
||||
| CBDD Single Insert mean | 5,169.8 μs | InsertBenchmarks CSV |
|
||||
| CBDD Batch Insert (1000) mean | 25,071.8 μs | InsertBenchmarks CSV |
|
||||
| CBDD FindById mean | 8.963 μs | ReadBenchmarks CSV |
|
||||
| CBDD FindById allocated | 36.23 KB | ReadBenchmarks CSV |
|
||||
| Offline compaction throughput | 9,073,346.22 bytes/sec | PerformanceGateSmoke JSON |
|
||||
| Offline compaction throughput | 944.26 pages/sec | PerformanceGateSmoke JSON |
|
||||
| Offline compaction throughput | 2,136.21 docs/sec | PerformanceGateSmoke JSON |
|
||||
| Compression OFF `Gen0/1/2` deltas | `30 / 4 / 2` | PerformanceGateSmoke JSON |
|
||||
| Compression ON `Gen0/1/2` deltas | `26 / 4 / 2` | PerformanceGateSmoke JSON |
|
||||
| Compression OFF allocated delta | 239,765,160 bytes | PerformanceGateSmoke JSON |
|
||||
| Compression ON allocated delta | 205,801,624 bytes | PerformanceGateSmoke JSON |
|
||||
|
||||
### Gate Outcome
|
||||
|
||||
- Write/read overhead is documented from fresh benchmark smoke artifacts.
|
||||
- Compaction throughput is documented with reproducible command and captured stats.
|
||||
- Compression-enabled workload does not show pathological GC escalation in this smoke scenario (Gen1/Gen2 unchanged, Gen0 and allocated bytes lower than compression-off).
|
||||
|
||||
|
||||
Reference in New Issue
Block a user