How proxy bandwidth is measured and how to plan your usage

What one gigabyte means here

BusyIP uses the binary gigabyte: 1 GB = 1 GiB = 1,073,741,824 bytes. The same unit is used in the traffic packs, account balance, and usage information.

When comparing costs, check the provider’s unit as well as its advertised price. A decimal gigabyte is a different number of bytes, so the label “GB” alone does not tell you the size of the allowance.

What the gateway counts

The meter counts traffic on the wire between your client and the BusyIP gateway. This includes traffic in both directions, TLS handshakes, and protocol framing. It is not limited to the final HTML, JSON, or file content your application saves.

Your application’s own counter may measure only the response body. That makes it useful for understanding your payload, but not directly equivalent to a counter measuring the entire connection.

Why overhead changes by workload

Opening a connection has a cost. If you transfer a substantial amount of content over it, that setup is a smaller share of the total. Many short transfers over fresh connections repeatedly pay that setup cost.

There is no single overhead percentage that accurately describes every workflow. The metering disclosure includes measurements from two different transfer sizes and explains why they differ.

Estimate with a representative sample

1. Record your account usage before a small test.
2. Run a representative batch of permitted requests, including the normal mix of page sizes and failures.
3. Record the usage after the test and count usable results.
4. Divide the traffic consumed by the number of usable results.
5. Multiply that observed traffic per result by the scale you are planning, then allow for variation.

For example, if your own sample uses 100 MiB for 200 usable results, the observed average is 0.5 MiB per result. This is an illustration of the calculation, not a BusyIP performance claim or a promised rate for your workload.

Reduce traffic you do not need

Reuse connections where your client and destination support it. When testing your own pages, decide whether images, fonts, and other resources are necessary to answer the question. Avoid repeated retries that cannot produce a useful result.

Keep efficiency changes tied to your test objective. Disabling resources makes a content check lighter, but it would undermine a test intended to verify that those resources actually load.

A balance that waits for your next job

Unused prepaid traffic does not expire. There is no recurring subscription or automatic recharge. A purchase adds bytes to the credential you already use.

Concurrent connection limits still apply: two for trial access and eight for paid access. Parallel requests do not turn a small traffic budget into a larger one; they only change how quickly it can be consumed.

Reconcile usage with the account tools

Use your account’s usage view to review bytes sent and received. When investigating a difference, compare the same time window and include all traffic your client made, including retries and setup.

The pricing page explains denied requests, bounded balance overshoot, and how those bytes appear after a top-up. If the numbers still do not make sense, contact support with the time window and a small reproducible example.