Self-Hosted Bandwidth and Data Transfer Calculator

Convert visits, pages, payload size, cache hit rate, and peak factor into monthly origin transfer and estimated average and peak throughput.

Default result: At the default workload, the origin serves 460 GB decimal per month, averaging about 1.40 Mbps with a modeled 14.0 Mbps peak.

Published and formula-checked 2026-09-27 · Runs locally in your browser · No input data is submitted

Origin per page
1.15 MB
Monthly origin transfer
460 GB
428.4 GiB
Average throughput
1.4 Mbps
Modeled peak
14 Mbps

Formula: origin MB/page = cacheable MB × (1 − cache-hit rate) + dynamic MB. Monthly transfer = visits × pages/visit × origin MB/page. Throughput uses an average 30.4375-day month.

Table of Contents

Transfer vs. throughput

Monthly transfer and peak throughput are different constraints. Monthly transfer is the total data your origin sends during a billing period. It helps you assess egress volume and a hosting allowance. Throughput is the rate at a moment, measured in bits per second. It helps you choose a home uplink, VPS network port, or other connection that can handle a burst.

A site can use a modest amount of data across a month but need a faster connection for a short rush of requests. The reverse is also possible: a steady stream can consume substantial transfer without a dramatic instantaneous peak. Do not use a monthly allowance as if it were a port-speed guarantee. This calculator estimates the origin path after the assumed CDN cache-hit rate, not every byte delivered to visitors. AWS defines bandwidth as a bit rate and uses decimal Mbps.

Exact formula

First calculate origin bytes for one page view:

Origin MB/page = cacheable MB/page × (1 − CDN cache-hit rate) + uncached MB/page

Then calculate monthly origin transfer:

Monthly origin MB = visits/month × pages/visit × origin MB/page

Average origin throughput is monthly megabytes converted to megabits and divided by the seconds in an average month:

Average Mbps = monthly origin MB × 8 / (30.4375 × 24 × 60 × 60)

The modeled peak applies the selected factor:

Modeled peak Mbps = average Mbps × peak factor

The 30.4375-day month keeps estimates comparable across calendar months. Divide decimal MB by 1,000 for decimal GB. GiB uses powers of 1,024, so keep units visible when comparing a result with a dashboard or allowance. This is a traffic model, not an invoice model; AWS transfer-modeling guidance recommends identifying traffic sources, destinations, and paths before estimating transfer.

Default worked example

The default inputs are 100,000 visits per month, 4 pages per visit, 3 MB of cacheable bytes per page, a 70% CDN cache-hit rate, 0.25 MB of uncached or dynamic bytes, and a peak factor of 10.

The origin receives cache misses plus dynamic bytes:

3 × (1 − 0.70) + 0.25 = 1.15 MB/page

Monthly origin transfer is:

100,000 × 4 × 1.15 = 460,000 MB

That equals 460 GB decimal, or about 428.4 GiB. Across an average month, the average origin throughput is about 1.40 Mbps. Applying the factor of 10 gives a modeled peak of about 14.0 Mbps.

QuantityDefault result
Origin bytes per page1.15 MB
Monthly origin transfer460 GB decimal
Binary equivalentAbout 428.4 GiB
Average origin throughputAbout 1.40 Mbps
Modeled peak throughputAbout 14.0 Mbps

These values describe the modeled origin path. They do not guarantee that a 10× burst will occur or that every visitor receives exactly 1.15 MB.

Cache-hit uncertainty

The cache-hit rate can dominate the estimate. A cache hit serves cacheable bytes at the edge; a miss sends those bytes to the origin. Therefore, lower hit rates increase origin transfer even when visits and pages per visit stay constant.

Never treat a CDN cache-hit assumption as guaranteed. Cookies, authorization, query strings, cache keys, revalidation, purges, geography, freshness, and personalization can change the effective rate. The rate may also vary by route, device, and time period.

Use a range when the rate is uncertain. With the default page bytes, a 90% hit rate gives 0.55 MB/page, while a 50% hit rate gives 1.75 MB/page. With other inputs fixed, monthly origin transfer changes in the same proportion. A measured range is more useful than a single optimistic value.

Peak factor

A peak factor turns an average rate into a modeled busy-period rate. A factor of 10 means the assumed peak is ten times the monthly average; it does not mean the site can sustain that rate all month or describe how long the burst lasts.

Use a higher factor for launches, campaigns, scheduled events, newsletters, or concentrated time zones. If the site is new, calculate several factors and choose the scenario that matters most. Leave headroom above the modeled peak because CPU, TLS, connections, storage, databases, queues, and CDN behavior can bottleneck before the link does.

For a home-hosted origin, upstream capacity is usually the key number: data leaving your connection toward visitors or a CDN. Residential service may advertise a large downstream rate but a smaller upload rate. Check upstream performance, sustained-use limits, data caps, carrier-grade NAT, and other restrictions that affect inbound hosting.

For a VPS or dedicated host, compare the network interface limit and monthly transfer allowance separately. A fast interface does not imply unlimited egress, and an allowance does not guarantee that the host can deliver the modeled peak without contention or other resource limits.

With a CDN, visitor delivery and origin delivery are separate paths. Size the origin for modeled origin traffic while considering the edge service's limits for total delivery. For deployment alternatives, see Vercel alternatives, Netlify alternatives, PaaS deployment options, and self-hosted PaaS.

Omitted traffic

This estimate covers the modeled page-view path only. It does not include:

  • Uploads, attachments, media ingestion, downloads, video, or large assets.
  • Backups, snapshots, replication, exports, or disaster-recovery copies.
  • Bots, crawlers, abuse, retries, health checks, or monitoring probes.
  • Protocol and encryption overhead, headers, connection setup, or retransmits.
  • APIs, background jobs, or cache misses from personalization, purges, or bypass rules.
  • Provider-specific billing, regional transfer, inter-zone traffic, or separately metered services.

Add relevant categories before choosing capacity. One backup schedule or large download can outweigh a page-view estimate.

Measuring real traffic

After launch, replace assumptions with origin and edge measurements. Record request count, response bytes, cache status, route, status code, region, and time bucket. Daily totals help validate monthly transfer; one-minute or five-minute buckets reveal peaks hidden by averages.

Compare origin egress with CDN analytics rather than inferring cache hit rate from visits. Investigate personalized routes, query-string variants, large assets, and purge windows. Separate users, bots, and operational traffic. Also confirm whether each report uses decimal GB, GiB, megabits, or bytes, and whether it counts data before or after compression.

Provider allowance comparison

Compare monthly origin transfer with the provider's documented allowance, matching units and scope. Separately compare modeled peak Mbps, plus headroom, with the advertised interface or uplink rate. Check whether inbound and outbound data differ, whether the allowance is per region or account, and whether service-to-service traffic is measured separately.

Do not choose from a price number alone. Plans can differ in included transfer, overage treatment, interface limits, measurement windows, and exclusions. AWS pricing principles identify compute, storage, and outbound transfer as major cost drivers and recommend verifying provider rates. Apply the same care to current allowance and metering documentation.

Next steps

Run the calculator with a conservative cache-hit range and at least two peak factors. Keep monthly transfer, average Mbps, and modeled peak Mbps as separate planning values. Add uploads, backups, bots, APIs, and other omitted categories that apply to your workload.

Before committing to a home uplink or host, verify upstream capacity, transfer measurement, interface limits, and failure behavior. After deployment, export edge and origin metrics regularly and revise the inputs when page weight, pages per visit, cache rules, or traffic concentration changes.

Frequently Asked Questions

What is the difference between GB and GiB?

GB is decimal: 1 GB equals 1,000 MB. GiB is binary: 1 GiB equals 1,024 MiB. Tools may use either unit. The default is 460 GB decimal, about 428.4 GiB, so match the calculator's unit to the allowance or dashboard.

Why does the Mbps result look low?

Monthly transfer is spread across every second of an average month, so its average rate can be modest. The default example is about 1.40 Mbps average and 14.0 Mbps at a 10× modeled peak. Short spikes or application limits may still require more headroom.

Does a CDN eliminate origin transfer?

No. Misses, dynamic responses, purges, revalidation, personalization, and bypass rules still create origin traffic. Treat cache hit rate as a measured planning assumption, never a guarantee. Visitor-facing delivery can remain substantial even when origin egress falls.

Will this estimate predict my provider bill?

No. It estimates modeled origin transfer and throughput. It excludes several traffic categories and does not reproduce provider-specific metering, allowances, regions, overages, or service-to-service rules. Use it to frame capacity questions, then compare with current provider documentation and measured traffic.

Calculator FAQ

What is the difference between GB and GiB?

This tool uses decimal GB for transfer totals because network providers commonly use base-10 units. It also shows GiB so storage-oriented comparisons remain clear.

Why can the average Mbps result look small?

Monthly transfer is spread across every second in the month. Real traffic is bursty, which is why the separate peak-factor input matters.

Does a CDN eliminate origin transfer?

No. Cache misses, dynamic responses, personalized pages, revalidation, and uploads still reach the origin. Treat cache hit rate as a measured assumption.

Will this estimate predict my hosting bill?

No. It estimates traffic. Providers differ on included transfer, regional rates, request charges, and overage rules, so apply the result to current provider terms.