Backup WAN Link Sizing Calculator
Find the minimum backup internet or WAN download and upload speed required during an outage. The estimate includes SD-WAN or VPN overhead, QoS priorities, traffic shedding, operational headroom, topology, and future growth.
Inputs
Advanced topology and primary-link context
Active-active multiplies the entered surviving-link rate. Shared N+1 divides the backup pool fairly across simultaneous contenders.
Optional context. The calculator compares class totals with capacity × measured peak utilization and reports backup capacity as a percentage of primary.
Advanced capacity assumptions
Measured sustained transport capacity ÷ provider rate; example: 510 Mbps sustained on 600 Mbps = 85%. Do not include tunnel expansion here.
Wire-rate expansion applied to payload; example: 100 Mbps payload at 8% becomes 108 Mbps. Includes VPN/SD-WAN headers.
Capacity intentionally unused after failover; example: 25% leaves 75% of sustained capacity for planned load.
Scales measured or estimated demand before overhead; example: 1.15 adds 15% for peak variation.
Growth, outage duration, and metered data
Growth is separate from headroom and compounds annually. Data uses decimal units: 1 GB = 8,000 megabits, assuming the retained rate is sustained for the outage.
Results
| Priority class | Direction | Desired wire load | Allocated | Action |
|---|---|---|---|---|
| Run a calculation to see traffic allocation. | ||||
Edit the three nominal download/upload candidates. Each policy uses the corresponding retained percentages shown below.
| Candidate | Download Mbps | Upload Mbps | Full service | Tiered failover | Critical-only |
|---|
| Application | Active endpoints | Down per endpoint | Up per endpoint | Contribution |
|---|---|---|---|---|
| Switch to Quick estimate to see application assumptions. | ||||
Advertisement
How to size a backup WAN link
- Choose an input mode. Use Quick estimate when monitoring data is unavailable, or Measured traffic for directional busy-hour or 95th-percentile observations.
- Separate traffic by direction and priority. Model download and upload independently, with critical and real-time traffic ahead of best effort.
- Choose retained percentages. Retained traffic should match the QoS, firewall, SD-WAN, or routing policy you can actually enforce.
- Set topology and capacity factors. Define sustained transport efficiency, encapsulation expansion, a peak safety multiplier, and operational headroom.
- Compare thresholds and policies. Review minimum viable and recommended rates, limiting direction, candidate circuits, growth, data allowance, and required shaping.
- Test before relying on it. A controlled failover test should verify path selection, throughput, latency, packet loss, and user-facing behavior.
Methodology and assumptions
The calculator treats WAN failover as a directional capacity and priority problem. For each direction it multiplies payload by the chosen retained percentage and peak safety multiplier, then applies encapsulation expansion. Minimum acceptable retention produces the minimum viable threshold. Desired retention divided by sustained transport efficiency and the capacity left after operational headroom produces the recommended nominal rate.
Units are decimal: 1 Mbps = 1,000,000 bits/s and 1 GB = 8,000 megabits. Efficiency describes measured sustained transport rate versus the provider rate; overhead describes added bytes on the wire; headroom is intentionally unused capacity after failover. Keeping these factors separate prevents double counting. The model does not simulate routing convergence, TCP slow start, packet loss recovery, application retry storms, SD-WAN path scoring, jitter buffers, cloud egress limits, carrier committed information rate, or device throughput caps. Treat the result as a planning estimate and validate it with device configuration, traffic monitoring, and failover drills.
Worked calculation using the measured defaults
The default symmetric example retains 80 Mbps of critical traffic, 36 Mbps of real-time traffic, 112 Mbps of business traffic, and 55 Mbps of bulk traffic: 283 Mbps of payload before the peak multiplier.
(80 + 36 + 112 + 55) × 1.15 × 1.08 = 351.49 Mbps desired wire load. A 600 Mbps link at 85% efficiency with 25% headroom has 600 × 0.85 × 0.75 = 382.50 Mbps planning capacity. The recommended nominal link is 351.49 ÷ (0.85 × 0.75) = 551.35 Mbps.| Class | Peak | Desired retained | Contribution before multiplier |
|---|---|---|---|
| Critical apps | 80 Mbps | 100% | 80 Mbps |
| Voice/video/real-time | 40 Mbps | 90% | 36 Mbps |
| Business apps | 160 Mbps | 70% | 112 Mbps |
| Best effort/bulk | 220 Mbps | 25% | 55 Mbps |
Starting bandwidth and link benchmarks
These are editable planning starts, not guarantees. Codec, resolution, content, packet size, RF conditions, and application behavior change real demand. Prefer directional 95th-percentile monitoring and a controlled failover test.
| Application | Starting estimate per active endpoint | Planning note and source |
|---|---|---|
| VoIP call | 0.10 Mbps down / 0.10 Mbps up | Allows more than the roughly 80 kbps IP rate for a G.711 stream; validate codec and packetization. Cisco voice design |
| HD video meeting | 2.5 Mbps down / 2.5 Mbps up | Microsoft lists recommended meeting video at 2.5 Mbps down / 4 Mbps up, with adaptive use; edit for your layout. Microsoft Teams |
| Cloud/SaaS active user | 0.50 Mbps down / 0.15 Mbps up | Generic interactive start only. File sync spikes with file size and change rate; pilot and measure. Microsoft OneDrive |
| Remote desktop | 1.50 Mbps down / 0.50 Mbps up | Microsoft gives 1.5 Mbps as a minimum for a light workload; resolution and frame rate raise demand. Microsoft RDS |
| CCTV camera/remote view | 0.25 Mbps down / 2 Mbps up | Direction depends on camera and recorder placement. Resolution, frame rate, codec, and motion dominate; measure the selected stream profile. Cisco video surveillance |
| Backup/bulk transfer | 50 Mbps down / 20 Mbps up per job | A policy-controlled starting cap, not an inherent requirement. Schedule, pause, or set an explicit bandwidth limit. AWS DataSync limits |
| Transport | Starting range | What to verify |
|---|---|---|
| Private fiber/Ethernet | 90–98% | Committed rate, handoff policer, device throughput |
| Business broadband | 75–90% | Contention, asymmetric upstream, sustained speed |
| IPsec/SD-WAN transport | 70–90% | Gateway encrypted throughput; enter packet expansion separately. RFC 9347 |
| LTE | 40–70% | Signal, cell load, shaping, antenna, data cap |
| 5G | 50–80% | Coverage layer, uplink, cell load, plan terms |
Capacity is not quality: verify round-trip latency, jitter, packet loss, queueing under load, and failover convergence time. Voice and interactive video can fail on a high-bandwidth path when delay variation or loss is poor.
Validation and provenance
No individual technical reviewer is claimed for this page. Before procurement, validate the result against:
- directional interface counters and carrier committed rates;
- NetFlow/IPFIX, firewall, or SD-WAN application analytics;
- VPN/SD-WAN appliance encrypted-throughput limits and MTU behavior;
- QoS shaping, blocking, and deferral policies during a controlled failover;
- latency, jitter, packet loss, failover time, cellular allowance, and billing terms.
FAQs
Must a backup WAN link match the primary link?
No. It should match the traffic that must survive the outage, not automatically the primary circuit rate. A smaller backup can be appropriate when enforceable QoS policy blocks or defers lower-priority traffic.
How do I obtain measured peak WAN traffic?
Use interface counters, NetFlow or IPFIX, firewall reports, SD-WAN analytics, or carrier graphs. Record download and upload separately and classify traffic during representative busy periods and a controlled failover test.
Should I use average traffic or the 95th percentile?
Use measured busy-hour peaks or a 95th-percentile rate rather than a monthly average. The 95th percentile filters rare spikes but still represents sustained demand; apply a separate peak safety multiplier for uncertainty.
How should I size an asymmetric backup link?
Calculate download and upload independently. Compare each directional requirement with the matching provider rate; the direction with the higher utilization or deficit constrains the design.
How do LTE or 5G data caps affect failover?
A cellular link may have enough instantaneous speed but insufficient allowance for a long outage. Estimate retained traffic over the expected outage duration and compare it with the plan allowance, carrier shaping rules, and billing terms.
What is the difference between bandwidth, latency, jitter, packet loss, and failover time?
Bandwidth is the transfer-rate ceiling; latency is delivery delay; jitter is variation in delay; packet loss is missing traffic; and failover time is how long routing takes to restore service. Adequate bandwidth alone does not guarantee application quality.
What do minimum viable and recommended capacity mean?
Minimum viable capacity carries each class only to its minimum acceptable percentage. Recommended capacity carries the desired retained percentages and preserves the selected operational headroom.
Is this backup WAN calculator private?
Yes. Calculations run locally in your browser and inputs are not uploaded. A share link places values in the URL only when you choose Share.
Disclaimer
Use these results for planning and comparison. Validate production WAN failover with network-device limits, carrier service terms, QoS policy, monitoring data, application testing, and documented recovery procedures.