Unmetered Dedicated Server Billing Explained: Metered vs Unmetered

Two dedicated servers with identical hardware and identical monthly traffic can produce completely different invoices depending on one line in the contract: whether bandwidth is billed as metered or unmetered. This is not a question of what "unlimited" technically means at the packet level — it is a billing and contract-structure question, and it is where most customers get surprised on their second or third invoice. This guide walks through exactly how providers price each model, the specific contract language to check before signing, and how to build a real cost model comparing the two for your actual traffic pattern.

Two Fundamentally Different Pricing Models

Metered billing charges you based on actual data transferred, usually with an included allowance and a per-GB or per-TB overage rate beyond it. Unmetered billing charges a flat rate for the port itself, regardless of how much data crosses it, up to the physical limit of the port speed and any fair-use ceiling buried in the acceptable-use policy. Both models can describe the exact same physical port and the exact same server — the difference is entirely in how the invoice is calculated, not in the underlying network hardware.

How Providers Actually Calculate a Metered Bill

A typical metered contract includes a base transfer allowance — commonly 10-30 TB per month on a 1Gbps port in 2026 — with overage billed per TB beyond that. Overage rates commonly range from $0.50 to $3 per TB depending on the provider's own upstream transit costs and the data center region (Asia-Pacific and South American regions often carry higher per-TB overage than North American or Western European regions due to transit routing costs).

Providers calculate your bill using one of two measurement conventions, and the difference matters:

• Total transfer (95th percentile is NOT used here) — simple sum of all inbound plus outbound bytes across the billing period, compared against your included allowance.
• 95th percentile billing — more common for larger commercial contracts, this measures your bandwidth utilization in 5-minute samples across the month, discards the top 5% of samples (to absorb short spikes), and bills based on the highest remaining sample. This can produce a lower bill than raw total-transfer metering if your traffic is spiky rather than sustained, since brief spikes get excluded from the calculation.

Always ask which convention applies to your contract — 95th percentile billing is far more forgiving of traffic bursts than raw total-transfer metering, and providers rarely volunteer which one they use unless asked directly.

How Providers Actually Structure Unmetered Contracts

An unmetered contract is priced around the port speed itself, not projected usage, because the provider is effectively selling you guaranteed access to that port's capacity rather than billing consumption. The provider's own cost model assumes a typical customer will not sustain anywhere near 100% utilization 24/7 — this is the same statistical oversubscription logic used across shared infrastructure generally. Fair-use policies exist specifically to catch the small percentage of customers who would sustain near-line-rate usage constantly, which would break the provider's own cost assumptions for that pricing tier.

The contract clause to look for is the specific fair-use threshold, expressed either as a sustained utilization percentage (e.g., "sustained usage should not exceed 95% of port capacity for more than X consecutive hours") or, less commonly, as an actual soft TB ceiling that triggers a conversation rather than an automatic bill. If a contract does not specify any threshold at all, that is a red flag — it usually means enforcement is discretionary and could be applied inconsistently.

Pricing Tiers: Metered vs Unmetered Across Port Speeds

Note that per-TB overage rates on higher port speeds are often lower than on smaller ports, since customers buying larger commitments typically negotiate better bulk transit pricing with the provider — a 10Gbps metered contract's overage rate is frequently cheaper per TB than a 1Gbps contract's overage rate, even from the same provider, which is easy to miss if you are only comparing the headline included allowance.

Special Cases: CDN Offload and Traffic Shaping

Many metered and even some unmetered contracts distinguish between traffic types in ways that are easy to overlook. Traffic served through a CDN (where the CDN, not your origin server, absorbs the bulk of end-user requests) counts differently than direct origin traffic — only cache misses and CDN-to-origin refresh traffic hit your billed port, which can dramatically reduce your effective billed transfer even on a high-traffic site. Some providers also offer discounted or free internal/private network transfer between your own servers within the same data center or region, which matters if you are running a multi-server architecture with heavy internal replication, backup, or database traffic that would otherwise count against your public transfer allowance. Always ask specifically whether private/internal network transfer is billed the same as public internet transfer — the answer varies enough between providers that it can materially change your total cost model.

Building a Real Cost Model: A Worked Example

Consider a server currently transferring 45 TB per month on average, with occasional spikes to 70 TB during promotional periods.

In this example, unmetered billing gives budget predictability and protects against the promotional spike months, but only makes financial sense if the flat unmetered rate is lower than what metered billing would cost you across a typical year including your spike months — run the actual numbers using your own transfer logs rather than assuming unmetered is always cheaper.

Contract Clauses to Read Before Signing

The Fair-Use Threshold

Get the exact sustained-usage percentage or duration threshold in writing. "Reasonable use" with no defined number is not a real answer — ask the sales team to point you to the specific clause in the acceptable-use policy document, not just their verbal assurance.

What Happens at the Threshold

Does the provider throttle your port speed, charge an overage fee despite the "unmetered" label, or terminate the contract with notice? Each of these has very different practical consequences, and providers vary significantly on this.

Burst Allowances

Some contracts (both metered and unmetered) include explicit burst allowances — short windows where you can exceed normal limits without penalty, useful for planned traffic spikes like product launches or livestreamed events.

Overage Billing Granularity

Confirm whether overages are billed monthly in arrears, require pre-purchase of additional TB blocks, or auto-scale with an approval step. Some providers require manual approval for any overage charge, which can mean unexpected throttling if you are unreachable during a spike.

Mid-Contract Rate Changes

Check whether the provider can change the per-TB overage rate or the fair-use threshold during your contract term, and what notice period applies if so.

Private Network and Internal Transfer Treatment

Confirm whether traffic between your own servers on the same internal network is billed the same as public transfer, since this can be a significant hidden cost for multi-server architectures with heavy replication or backup traffic if it is not called out separately.

CDN and Cache-Miss Traffic Definitions

If you run behind a CDN, ask whether the contract distinguishes CDN-origin traffic from direct visitor traffic, since this materially changes your effective billed transfer on high-traffic sites that get most of their requests served from CDN cache rather than your origin server directly.

How to Read Your Own Bandwidth Graphs Correctly

Before modeling any cost comparison, pull your provider's bandwidth graphs and look at three things separately: average sustained utilization, peak 95th-percentile utilization, and total monthly transfer in TB. These numbers tell different stories — a server averaging 15% utilization with occasional spikes to 80% during a weekly batch job has a very different cost profile under metered versus unmetered billing than a server steadily running at 40% utilization around the clock, even if both end up transferring similar total TB in a typical month. Most control panels (WHM, Plesk, or a dedicated monitoring stack like Zabbix or Grafana with an SNMP or netdata data source) can show these three views separately; if your current provider only shows you a single "total bandwidth used" number without a time-series graph, ask for one before trying to model your real costs, since the shape of your traffic over time matters as much as the total.

When Metered Billing Is Actually the Better Deal

• Your traffic is well below any reasonable unmetered fair-use ceiling and consistently so — you are effectively subsidizing headroom you never use.
• Your traffic is highly seasonal with long quiet periods, where a metered plan's lower base cost during quiet months outweighs occasional overage charges during peak months.
• You are running a small number of low-traffic sites where predictability matters less than minimizing baseline monthly cost.

When Unmetered Billing Is Actually the Better Deal

• Your traffic is already trending toward or above a typical metered allowance most months.
• You need predictable, fixed hosting costs for internal budgeting or client billing purposes.
• Your traffic pattern includes occasional large spikes (product launches, viral content, backup windows) that would trigger expensive overage charges under a metered model.
• You are reselling hosting or billing clients a flat monthly rate yourself, where an unpredictable underlying cost from metered overages would erode your own margin unexpectedly.
• Your team does not have the bandwidth to actively monitor usage and respond to approaching thresholds, making a flat unmetered rate operationally simpler even at a modest cost premium.

A Note on Reseller and Agency Billing Models

If you are hosting multiple client sites on a single dedicated server and billing those clients yourself, the metered-vs-unmetered decision carries an extra layer of complexity: you need predictability not just for your own budgeting but to avoid a scenario where one client's traffic spike (a viral post, a poorly optimized plugin causing runaway API calls, or even a compromised site being used to send spam or serve malware) blows through your allowance and triggers overage charges you cannot easily pass back to that specific client after the fact. Many resellers default to unmetered billing specifically to insulate their own margin from this risk, treating the flat-rate premium as a form of insurance rather than evaluating it purely on projected average traffic. If you do take this approach, it is still worth monitoring per-client bandwidth usage internally (most control panels support this) so you can have an informed conversation with a client whose site is disproportionately driving your infrastructure costs, even if your own billing to the provider is unaffected.

Seasonal and Growth-Stage Businesses: A Special Consideration

E-commerce operations with sharp seasonal peaks (holiday shopping periods driving 3-5x normal traffic for a few weeks a year) face a genuinely harder version of this decision, since neither model is obviously right across the full year. A metered plan sized for average traffic gets hit with painful overage charges during peak weeks, while an unmetered plan sized to comfortably absorb peak-week traffic means paying for headroom that sits unused for 45+ weeks of the year. Some providers offer a middle path here — a metered plan with a pre-purchased "burst pack" of additional TB you can activate for a defined period around your known peak season, which avoids both the year-round unmetered premium and the risk of uncontrolled overage billing during your highest-traffic weeks. If your business has a genuinely predictable seasonal pattern, it is worth asking your provider directly whether this kind of temporary burst allowance exists before defaulting to either standard billing model.

Buyer's Checklist Before Choosing a Billing Model

• Pull at least 3-6 months of your actual bandwidth logs, not a guess, before comparing plans.
• Ask for the exact fair-use threshold on any "unmetered" plan in writing, including what enforcement action follows.
• Confirm whether metered billing uses total-transfer or 95th-percentile measurement — this changes the math significantly for spiky traffic.
• Model your cost under both billing structures using your real traffic data, including your highest historical spike month.
• Ask what happens contractually if you exceed limits — throttling, overage billing, or termination — before signing, not after.
• Re-evaluate annually — traffic patterns that favored metered billing at signup can shift enough within a year to make unmetered the better deal, or vice versa.

Frequently Asked Questions

Is unmetered the same as unlimited bandwidth?

Not contractually. "Unmetered" describes the billing method — you are not billed per GB — while "unlimited" is a marketing term that still typically operates within a fair-use policy. The two terms are often used interchangeably in sales copy, but the billing mechanics are what actually matter, and that is defined in the fair-use clause, not the label used on the pricing page.

Can a provider switch me from unmetered to metered billing mid-contract?

Only if your contract explicitly allows it, which is why checking the mid-contract rate change clause before signing matters. Reputable providers lock your billing model for the contract term and only revisit it at renewal.

What is 95th percentile billing and why does it matter?

It is a metered-billing measurement method that samples your bandwidth every 5 minutes, discards the top 5% of samples, and bills based on the highest remaining sample rather than total data transferred. It is generally more forgiving of short traffic spikes than simple total-transfer metering, so it matters a lot if your traffic is bursty rather than steady.

Why do some providers not disclose their fair-use threshold publicly?

Providers sometimes keep the exact threshold vague to preserve flexibility in how they handle edge cases, and in some instances simply because their own capacity planning assumptions are proprietary. This is exactly why it is worth asking directly and getting the specific number or percentage in writing during the sales process rather than accepting "reasonable use" as an answer, since a written commitment gives you real recourse if a dispute arises later.

Do unmetered plans ever have a hidden TB cap?

Some do, expressed as a fair-use ceiling rather than a hard cap, though enforcement varies. Always ask for the specific number or percentage rather than accepting "fair use" as a vague answer.

How do I estimate my future bandwidth needs if my traffic is still growing quickly?

Plot at least 6-12 months of historical transfer data on a simple growth curve and extrapolate forward, then add a buffer for unplanned spikes (a successful marketing campaign, a press mention, a viral social post) rather than assuming linear, predictable growth. Revisit the model every quarter for a fast-growing business, since a billing structure that was correctly chosen at signup can become the wrong choice within two or three quarters of sustained growth.

Which billing model is cheaper for a small blog with steady low traffic?

Metered billing is usually cheaper for genuinely low, steady traffic, since you are not paying a flat rate for port capacity you will never use. The crossover point depends on the specific provider's pricing, which is why running the numbers on your own traffic logs matters more than a general rule.

Should I negotiate my billing model at contract renewal?

Yes — traffic patterns change, and renewal is the natural point to re-run your cost model and switch billing structures if your usage has shifted meaningfully since you first signed.

Does CDN traffic count against my dedicated server's bandwidth allowance?

Only the traffic between the CDN and your origin server (cache misses and periodic refreshes) counts against your server's billed transfer — traffic the CDN serves directly to end users from its own cache does not touch your port at all. This is one of the most effective ways to reduce billed transfer on a metered plan without changing your billing model, simply by putting a CDN in front of static and cacheable content.

Are overage charges billed immediately or at the end of the billing cycle?

This varies by provider — some bill overages in arrears at the end of the monthly cycle alongside your base fee, while others require pre-purchasing additional TB blocks once you approach your allowance, with a hold or service pause if you do not respond in time. Ask specifically how overage billing timing works, since a provider that pauses service pending payment approval can create unplanned downtime during a legitimate traffic spike.

Does the billing model affect DDoS mitigation or security features?

Not directly — billing model and mitigation capability are usually separate product decisions, though very large volumetric attacks can occasionally trigger fair-use conversations on unmetered plans if mitigated traffic still counts toward the port's sustained utilization. Confirm explicitly whether DDoS-related traffic (both the attack itself and the mitigation overhead) is excluded from fair-use calculations before an incident happens, not during one.

Can I switch billing models without changing servers?

Often yes, if the provider offers both models on the same hardware tier — this is usually a straightforward account-level change rather than a migration. Confirm with your account manager whether a mid-contract switch is possible and whether it resets any existing contract term commitment.

WebsNP offers both metered and unmetered billing structures across our dedicated server plans, and our team can review your actual bandwidth logs to recommend the model that fits your real traffic pattern. Contact us for a billing review, or see current pricing tiers.