Bandwidth Calculator

Port speed and data transfer are two different things that get quoted in the same breath, and the gap between them trips up almost every server order. A port is a rate (bits per second) while transfer is a volume (bytes over a month). This calculator converts between the two in both directions: give it a port speed and utilization and it returns the monthly transfer; give it a transfer target and it returns the sustained port speed required to hit it.

The anchor number to keep in your head: a single 1 Gbps port, running flat out for a 30-day month, moves 324 TB. Nothing real runs flat out (the utilization slider is where you model that), but every other figure on this page is that one scaled up, down, or inverted. Start from 324 TB and the rest of the arithmetic follows.

Turning a port speed into monthly transfer

The conversion is four multiplications and one unit correction. Start with the line rate in bits per second: 1 Gbps is 1,000,000,000 bits/s. Divide by 8 to get bytes per second, because links are sold in bits and transfer is counted in bytes; this ÷8 is where most back-of-envelope estimates go wrong by a factor of eight. Multiply by 86,400 seconds in a day, then by the number of days in the billing period, then by your utilization fraction. Divide by 10¹² to land in decimal terabytes.

Worked through at 100% utilization over 30 days: 1 Gbps ÷ 8 × 86,400 × 30 ÷ 10¹² = 324 TB. A 10 Gbps port is exactly ten times that: 3,240 TB, or about 3.24 PB. Drop utilization to a more realistic 50% and both halve: 162 TB on the 1 Gbps port, 1,620 TB on the 10 Gbps. Utilization scales the answer linearly, so a port you keep at 20% average moves a fifth of its theoretical ceiling (64.8 TB on 1 Gbps), which is why the sustained-average number matters far more than the port label when you are sizing transfer.

The calculator also reports the same volume as tebibytes (TiB), per-day TB, per-hour GB, and the effective sustained rate in Mbps. The TiB figure exists because operating systems and many monitoring tools count in binary units (1 TB is about 0.9095 TiB), so 324 TB shows up as roughly 294.68 TiB once your OS reports it.

How that transfer becomes a bill

Three models dominate, and the same traffic can cost wildly different amounts depending on which one you are on. The mechanics of each are worth knowing precisely, because the marketing rarely spells them out.

95th-percentile billing charges for sustained rate, not total volume. The provider samples the traffic on your port every 5 minutes, in both directions, for the whole month: roughly 8,640 samples. At the end of the month those samples are sorted from lowest to highest, the top 5% (about 432 samples, or ~36 hours) are discarded, and you are billed on the highest sample that remains, the 95th-percentile rate. The effect is that you get roughly a day and a half of unbilled bursting: short spikes fall into the discarded 5% and cost nothing, while anything you sustain shows up in the bill.

Metered billing charges for total volume: a price per TB moved, measured against the transfer number this calculator produces. Unmetered (flat-port) billing charges a fixed price for the port itself and does not count bytes at all; you pay the same whether the port sits idle or runs at the line rate you provisioned.

• 95th percentile. Favours spiky, bursty traffic (CDN edges, occasional backups): the peaks you sustain briefly are discarded, and you pay for the rate you hold, not the total moved.
• Metered per-TB. Favours low, predictable volume: cheap when you move little, but scales directly with every terabyte, so it punishes sustained transfer.
• Unmetered flat-port. Favours high, sustained utilization: the cost is fixed, so the more of the port you actually use, the lower your effective cost per TB.

Sizing the port to the workload

Size against your average, then check the peak. Most workloads spend most of the month well below their busiest hour, so the sustained-average utilization (not the momentary spike) is what determines the transfer volume and, on metered or unmetered plans, the sensible port size. Run the calculator at your realistic average utilization to get the transfer figure, then run it again at your peak utilization to confirm the port has the headroom to serve that peak without saturating.

The rule of thumb worth internalizing: if your sustained average utilization is creeping past about 50% of the port, you are close to the point where the next port size up, or an unmetered plan, is the better call. A link held near saturation has no headroom for growth or spikes, and on 95th-percentile billing a consistently high sustained rate is exactly the expensive case; the bursts you hoped were free are now the baseline. Sustained demand wants a bigger pipe or a flat port; occasional bursts on an otherwise-quiet link are what 95th-percentile pricing was built to reward.