Model
The queueing wall (M/M/1)
For one server with random arrivals and random service times, the average response time.
- Response time
- average response time (waiting plus service)
- Service time
- the average service time
- Utilization
- fraction of time the server is busy = arrival rate × average service time
How it behaves
One input on the horizontal axis, the others held fixed.
Data table
| Utilization ρ | Response time R (× S) |
|---|---|
| 0 | 1× |
| 0.012 | 1.01× |
| 0.025 | 1.03× |
| 0.037 | 1.04× |
| 0.049 | 1.05× |
| 0.061 | 1.07× |
| 0.074 | 1.08× |
| 0.086 | 1.09× |
| 0.098 | 1.11× |
| 0.11 | 1.12× |
| 0.123 | 1.14× |
| 0.135 | 1.16× |
| 0.147 | 1.17× |
| 0.159 | 1.19× |
| 0.171 | 1.21× |
| 0.184 | 1.23× |
| 0.196 | 1.24× |
| 0.208 | 1.26× |
| 0.221 | 1.28× |
| 0.233 | 1.3× |
| 0.245 | 1.32× |
| 0.257 | 1.35× |
| 0.269 | 1.37× |
| 0.282 | 1.39× |
| 0.294 | 1.42× |
| 0.306 | 1.44× |
| 0.319 | 1.47× |
| 0.331 | 1.49× |
| 0.343 | 1.52× |
| 0.355 | 1.55× |
| 0.367 | 1.58× |
| 0.38 | 1.61× |
| 0.392 | 1.64× |
| 0.404 | 1.68× |
| 0.416 | 1.71× |
| 0.429 | 1.75× |
| 0.441 | 1.79× |
| 0.453 | 1.83× |
| 0.466 | 1.87× |
| 0.478 | 1.91× |
| 0.49 | 1.96× |
| 0.502 | 2.01× |
| 0.514 | 2.06× |
| 0.527 | 2.11× |
| 0.539 | 2.17× |
| 0.551 | 2.23× |
| 0.564 | 2.29× |
| 0.576 | 2.36× |
| 0.588 | 2.43× |
| 0.6 | 2.5× |
| 0.613 | 2.58× |
| 0.625 | 2.66× |
| 0.637 | 2.75× |
| 0.649 | 2.85× |
| 0.661 | 2.95× |
| 0.674 | 3.07× |
| 0.686 | 3.18× |
| 0.698 | 3.31× |
| 0.71 | 3.45× |
| 0.723 | 3.61× |
| 0.735 | 3.77× |
| 0.747 | 3.96× |
| 0.759 | 4.16× |
| 0.772 | 4.38× |
| 0.784 | 4.63× |
| 0.796 | 4.91× |
| 0.808 | 5.22× |
| 0.821 | 5.58× |
| 0.833 | 5.99× |
| 0.845 | 6.46× |
| 0.857 | 7.02× |
| 0.87 | 7.68× |
| 0.882 | 8.47× |
| 0.894 | 9.46× |
| 0.906 | 10.7× |
| 0.919 | 12.3× |
| 0.931 | 14.5× |
| 0.943 | 17.6× |
| 0.956 | 22.5× |
| 0.968 | 31× |
| 0.98 | 50× |
Worked examples, recomputed
Every number the mechanisms quote for this relationship is recomputed from it, so none of them disagrees with it.
- speculation lifts ρ from 0.90 to 0.95 → the queue factor goes from 10 to 20
- four spillers at 0.3 GB/s on a 1.5 GB/s drive: ρ = 0.8 → 5×
- ρ = 0.5 → 2 × S
- ρ = 0.8 → 5 × S
- ρ = 0.9 → 10 × S
- ρ = 0.95 → 20 × S
- ρ = 0.99 → 100 × S
Where it appears
Each mechanism states it in its own names. The general symbol each name stands for is shown with it.
- Physical Substrate & Performance Laws
Why do trade-offs exist at all?
- Physical Substrate & Performance Laws
Why do trade-offs exist at all?