ISP contention and capacity planner
How much room is left on your upstream link? Compare 10, 20 and 50 Mb/s plans with evening demand, then model the next 100 subscribers.
Model your evening peak
All calculations run in this browser. Nothing you enter is submitted or saved. Units are downstream Mb/s (megabits per second).
Planning results
- Sold bandwidth / capacity
- 7.5:1
- Expected busy-hour demand
- 450 Mb/s
- Expected capacity use
- 90%
- Planning requirement with minimum
- 530 Mb/s
200 subscribers; 80 expected active at peak. Sold plan speeds total 3,750 Mb/s.
Capacity shortfall: 30 Mb/s. This is an aggregate estimate, not a speed guarantee or a safe sales limit.
Adding the next 100 subscribers
At the same plan proportions: 795 Mb/s total required, or 295 Mb/s extra above current capacity.
| Plan | More subscribers | Total for +100 | Extra capacity |
|---|---|---|---|
| 1: 10 Mb/s | 0 | 730 Mb/s | 230 Mb/s |
| 2: 20 Mb/s | 0 | 770 Mb/s | 270 Mb/s |
| 3: 50 Mb/s | 0 | 1,130 Mb/s | 630 Mb/s |
Assumptions you can change
These defaults are an illustrative scenario, not industry averages or recommended contention settings. Replace them with measurements from your own evening peak.
- 500 Mb/s usable upstream: a round example of a shared transit or backhaul bottleneck. Enter usable downstream capacity after protocol overhead and your chosen operating reserve, not the port label. No extra safety margin is added.
- 100 × 10, 75 × 20 and 25 × 50 Mb/s plans: an example of 200 subscribers across three tiers, chosen to make plan mix visible. A zero-count row lets you test a new tier.
- 40% concurrency: assumes 80 of the 200 subscribers are active together. This is the share consuming traffic during the busy hour, not the share with a connected session.
- 30% utilisation per active subscriber: assumes an active user consumes 30% of their plan rate on average. Applied equally to every tier for a simple model; measure each tier separately if their usage differs.
- 5 Mb/s minimum per active user: a chosen planning floor to illustrate why average demand can understate the desired capacity. It must not exceed any entered plan speed. Set it to zero to model average demand alone.
- Next 100 subscribers: keeps the existing subscriber proportions, allowing fractional counts as a weighted forecast. With the example mix, it adds 50, 37.5 and 12.5 subscribers. The per-plan table also shows 100 additions to one tier at a time.
Formulas, including without JavaScript
For each plan i, let Sᵢ be its speed in Mb/s and Nᵢ its subscriber count. Let C be usable capacity, c the concurrency percentage divided by 100, u the utilisation percentage divided by 100, and T the target minimum speed.
- Contention ratio: Σ(Sᵢ × Nᵢ) / C. Undefined when capacity is zero.
- Expected busy-hour demand D: Σ(Sᵢ × Nᵢ × c × u). Expected capacity use is 100 × D / C.
- Planning requirement P: Σ(Nᵢ × c × max(Sᵢ × u, T)). Each plan reserves the larger of expected active-user demand and the chosen floor.
- Headroom for plan i: max(0, floor((C − P) / (c × max(Sᵢ × u, T)))). No finite estimate when the denominator is zero. Each plan gets a separate scenario using all remaining headroom.
- Total capacity for 100 more, same mix: P × (N + 100) / N, where N = ΣNᵢ. Undefined with no existing subscribers. For one plan only: P + 100 × c × max(Sᵢ × u, T).
- Extra capacity to buy: max(0, total required − C). This also includes any current shortfall.
Worked example: 3,750 Mb/s sold ÷ 500 Mb/s = 7.5:1. Expected demand is 450 Mb/s (90% of capacity). The planning floor raises the requirement to 530 Mb/s: a 30 Mb/s shortfall and zero additional subscriber headroom. Keeping the same mix for 100 more subscribers needs 795 Mb/s total, or 295 Mb/s extra.
Before buying capacity or selling more plans
This is one aggregate downstream bottleneck with the same concurrency and utilisation assumptions across all plans. It does not simulate upload demand, bursts, packet loss, latency, queue scheduling, radio airtime, sector limits, backhaul topology, outages or failover. Run separate scenarios for each real bottleneck and direction. Fractional active users represent an average, not a guaranteed maximum.
Compare the estimate with measured busy-hour traffic, test higher concurrency (including 100%), and keep your own operating reserve. A lower contention ratio alone does not establish a better subscriber experience.
Common questions
What is an ISP contention ratio?
It is the sum of sold plan speeds divided by usable shared upstream capacity. A 7.5:1 ratio means 7.5 Mb/s of sold speed per 1 Mb/s of capacity. It does not describe actual simultaneous demand or guarantee a speed.
How many more subscribers can I sell?
For each plan, the planner divides remaining planning capacity by the assumed load per new subscriber and rounds down. Each row assumes only that plan grows. Validate the assumptions with busy-hour measurements and leave an operating margin before making sales decisions.
Does this minimum-speed target guarantee service quality?
No. It reserves aggregate capacity for the assumed active population, using the larger of average demand and the target for each plan. It cannot guarantee scheduling fairness, latency, packet loss or per-user throughput.
Does ISPAgents read my router or save these inputs?
No. This standalone tool calculates locally in your browser. It does not connect to your router, send the inputs, or save them between visits.
Plan the operations around subscriber growth
ISPAgents provides subscriber records and TR-069/CWMP device management. CRM and RouterOS operations are guided pilot foundations with scope agreed during onboarding.
Explore WISP operations or review TR-069 device management. Return to all ISP planning tools.
