Performance of atomic deployments using ssi atomic. Atomic deploys provide zero-downtime updates with automatic rollback.
📈 Performance History
Atomic Basic Deploy Performance History
xychart x-axis [0.244.0, 0.244.2, 0.244.10, 0.244.17, 0.244.19, 0.244.24, 0.244.25] y-axis "Time (ms)" 0 --> 8 line [6.13, 6.09, 6.47, 6.57, 6.07, 6.28, 6.26] line [5.84, 5.82, 6.06, 6.17, 5.83, 5.96, 5.94] line [5.56, 5.55, 5.66, 5.80, 5.59, 5.65, 5.62]
📊 Performance Results
| Test | Description | Mean (ms) | Median (ms) | ± Std Dev | CV% | Outliers | Range | Peak RSS (MB) | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Basic Site | High-iteration atomic deployment of the basic performance test site | 5.94 | 5.94 | ± 0.16 | 2.7% | 166 (16.6%) | 5.62-6.26 | 43.90 | 🟢🟢🟢 |
| vs Deploy (same site) | -0.11 ms | -0.11 ms | 0.22 ms | — | — | — | — | 🟢⚪⚪ | |
Methodology
- Outlier Detection: IQR method (Q1 - 1.5×IQR to Q3 + 1.5×IQR) removes system noise
- Statistics Reported: Cleaned data after outlier removal
- Primary Metrics: Mean (cleaned), Median (outlier-resistant), CV% (quality indicator)
- Measurement Method: Wall-clock timing with subprocess overhead
- Iterations: 1000 runs for atomic deploy (same as basic performance test)
Atomic deployments provide zero-downtime updates with automatic rollback on failure. These benchmarks help track the overhead of the atomic operation itself.