Quick Answer
48 × 1 Gbps access capacity = 48 Gbps
2 × 10 Gbps uplinks = 20 Gbps aggregate uplink
Theoretical oversubscription = aggregate access port rate ÷ aggregate uplink rate = 48 Gbps ÷ 20 Gbps = 2.4:1
Use the same traffic direction and capacity basis for both sides of the ratio. This is a port-rate comparison, not a promise of application throughput: protocol overhead, bursts and the switch forwarding capacity affect usable throughput. A 2.4:1 ratio is not automatically a problem; compare expected busy-hour demand with the 20 Gbps aggregate uplink.
Use the NetEngineerLab calculator to compare theoretical capacity, practical utilization and failure-state headroom.
Open the Calculator →When 2×10G Is Usually Enough
Two 10G uplinks are more likely to be sufficient when the switch serves ordinary user-access traffic and the endpoints rarely transmit near 1 Gbps at the same time.
Example:
Assumption: 40 ports are active in the busy hour, each access port is 1 Gbps, and their average simultaneous utilization is modeled at 25%.
Estimated offered load = active ports × access rate × busy-hour utilization
40 × 1 Gbps × 0.25 = 10 Gbps estimated demand
Against 20 Gbps aggregate uplink capacity, modeled headroom is 20 − 10 = 10 Gbps.
This is a planning estimate, not a measured result; use representative busy-hour counters and account for traffic direction and bursts when available.
| Modeled busy-hour case | Assumption | Estimated load | Headroom vs 20 Gbps |
|---|---|---|---|
| Light use | 20 active ports × 1 Gbps × 25% | 5 Gbps | 15 Gbps |
| Typical planning example | 40 active ports × 1 Gbps × 25% | 10 Gbps | 10 Gbps |
| Higher concurrency | 40 active ports × 1 Gbps × 50% | 20 Gbps | 0 Gbps before overhead or bursts |
These are illustrative assumptions, not universal utilization benchmarks. The higher-concurrency case reaches the 20 Gbps aggregate uplink capacity; measured demand, traffic mix and required reserve determine whether that operating point is acceptable.
When 2×10G May Not Be Enough
- High simultaneous east-west traffic
- Large backup or replication jobs
- Server or storage workloads
- High-throughput Wi-Fi aggregation
- Multigig access ports
- Rapid growth in endpoint count or traffic
- Strict N-1 capacity requirements
Allow for a Single Uplink Failure
With two active 10 Gbps members, losing one leaves 10 Gbps. For example, a modeled 12 Gbps busy-hour load fits within 20 Gbps in normal operation but exceeds the surviving 10 Gbps by 2 Gbps. This is the basic N-1 check; detailed failure-domain and surviving-capacity analysis belongs in the N-1 switch uplink guide.
LAG Does Not Give 20G to Every Flow
A 2×10G LAG can provide 20 Gbps in aggregate, but a single flow normally hashes to one member link. If the traffic pattern is dominated by a small number of large flows, the aggregate capacity may not be used evenly.
When to Consider an Uplink Upgrade
Consider faster or additional uplinks when measured busy-hour demand plus planned growth approaches the usable capacity your design requires, or when the surviving capacity after a member failure cannot carry the required load. Check that the switch, optics, cabling and peer support the proposed rate. Detailed 10G-versus-25G capacity and platform trade-offs belong in a dedicated uplink planning guide; this page does not assume a particular upgrade is universally better.
A Better Decision Rule
For normal operation, ask: Does the measured or transparently modeled busy-hour demand fit within the 20 Gbps aggregate capacity with the headroom this network requires? Then make a separate basic N-1 check for the surviving 10 Gbps member capacity.
Decision Checklist
- Measure representative busy-hour traffic in each relevant direction.
- Confirm both 10G links actively forward and whether they form a LAG.
- Compare demand plus planned growth with 20 Gbps aggregate capacity, allowing for headroom and equipment limits.
- Check whether 10 Gbps surviving capacity meets the required basic N-1 target.
- Review faster or additional uplinks only if the demand and resilience targets require them.
Frequently Asked Questions
What is the oversubscription ratio for 48×1G with 2×10G uplinks?
2.4:1.
Can 2×10G support 48 Gigabit users?
Often yes for typical user-access traffic, but not automatically. Busy-hour demand and failure-state requirements decide whether the design is sufficient.
What is the basic single-link failure check?
With one 10 Gbps member unavailable, compare expected busy-hour demand with the remaining 10 Gbps; detailed N-1 analysis is outside this guide.