Start Here
If you are new to the topic, begin with the main guide. It explains the formula, theoretical vs practical capacity, LAG behavior, N-1 failure and growth planning.
Calculate Your Design
Use the engineering calculator when you want to move from theory to a real switch design.
Model port count, port speed, uplink count, uplink speed and practical capacity assumptions.
Open the Calculator →Worked example: 48×1G access with 2×10G uplinks
Static engineering case you can re-enter in the switch uplink oversubscription calculator. Numbers match the N-1 guide examples on this topic.
1. Inputs and assumptions
| Input | Value | Assumption |
|---|---|---|
| Access ports × speed | 48 × 1 Gbps | All ports in the demand pool |
| Uplinks | 2 × 10 Gbps (LACP) | Members share load in the normal state |
| Busy-hour demand | 13 Gbps toward the uplink | Planning estimate, not line-rate on every port |
| Failure case | One 10G member down (N-1) | Remaining capacity = 10 Gbps |
2. Formula and units
Theoretical oversubscription = (access ports × access speed) ÷ (uplink count × uplink speed)
Normal headroom = uplink capacity − busy-hour demand
N-1 headroom = surviving uplink capacity − busy-hour demand
Units: Gbps throughout.
3. Calculation
| Step | Calculation | Result |
|---|---|---|
| Access pool | 48 × 1 | 48 Gbps |
| Normal uplink capacity | 2 × 10 | 20 Gbps |
| Theoretical ratio | 48 ÷ 20 | 2.4:1 |
| Normal headroom | 20 − 13 | +7 Gbps |
| N-1 capacity | 1 × 10 | 10 Gbps |
| N-1 headroom | 10 − 13 | −3 Gbps (deficit) |
4. Result and limits
Normal state can carry the 13 Gbps busy-hour estimate; the single-uplink failure state cannot. A 2×10G design for this demand needs either lower busy-hour traffic, a third/fourth 10G member, or 25G-class uplinks before approval.
Boundaries: ignores LAG hash imbalance, microbursts, and oversubscription inside the fabric. Not a substitute for measured counters or vendor QoS limits. Cross-check with the N-1 guide and the calculator.
Common Scenarios
Failure & Resilience
Oversubscription should not be approved using normal-state bandwidth alone. A resilient design should also be checked after the loss of an uplink or other required N-1 component.
- Normal-state oversubscription
- N-1 surviving uplink capacity
- Busy-hour demand under failure
- Growth margin after failure
Calculate surviving uplink capacity, headroom and upgrade options.
Engineering Resources
The GitHub versions are concise engineering references intended for quick lookup and reuse.
Recommended Learning Path
- Read the main oversubscription guide.
- Work through the 48-port switch example.
- Use the calculator with your own port and uplink values.
- Repeat the calculation under N-1.
- Compare realistic upgrade options such as 2×10G, 4×10G and 2×25G.
Current Topic Assets
| Asset | Purpose | Status |
|---|---|---|
| Main Guide | Complete oversubscription concept and workflow | Available |
| 48-Port Example | Focused practical scenario | Available |
| Calculator | Interactive capacity planning | Available |
| GitHub Cheat Sheet | Concise engineering reference | Project file ready |
| GitHub 48-Port Example | Focused Markdown example | Project file ready |
| 2×10G Problem Page | Long-tail search question | Available |
| LACP Member Congestion Guide | Member-level traffic diagnosis and N-1 example | Available |
| N-1 Problem Page | Failure-state capacity analysis | Available |