What is Bandwidth?

How Bandwidth Works

Every link in a network path has a capacity — the maximum bits per second it can carry. End-to-end throughput is set by the narrowest link on the path, so in a proxied request the client's connection, the proxy's connection, and the target server's capacity all matter; whichever is smallest becomes the ceiling. Units invite confusion: network capacity is quoted in megabits per second (Mbps), file sizes in megabytes (MB), and there are eight bits to a byte, so a 100 Mbps link tops out around 12.5 MB per second of payload.

Actual throughput usually lands below the ceiling. TCP needs time to ramp up, and on high-latency paths a single connection may not keep enough data in flight to fill the pipe — the bandwidth-delay product problem. This is why parallel connections often achieve far more combined throughput than one connection, especially through residential or mobile exits.

Capacity can also be shared or dedicated. Residential exits ride real household connections whose speed varies by device and time of day, while dedicated allocations reserve a fixed rate for a single customer.

Why It Matters for Proxies and Scraping

Bandwidth is the axis proxy billing is built on. Metered plans charge per gigabyte transferred, so data-heavy jobs — rendered pages, images, video, large API payloads — cost proportionally more, while lightweight HTML fetching stays cheap; unmetered plans remove the volume cap and charge a flat rate. ProxyOmega's Budget Unlimited and Premium Unlimited plans are unmetered, while Platinum and Mobile bill per GB.

Volume planning prevents surprises. Estimate total transfer as pages multiplied by average page weight, and remember that rendering in a headless browser pulls scripts, images, and fonts that a plain HTTP fetch skips. Blocking unneeded asset types at the browser level routinely cuts transfer volume by a large margin on metered routes.

Practical Notes and Common Misconceptions

The phrase unlimited bandwidth in proxy marketing nearly always means unmetered data volume, not infinite speed — throughput is still bounded by the path's real capacity. And more bandwidth does not fix every slow job: if the bottleneck is latency, per-IP rate limits, or target-side throttling, a bigger pipe changes nothing.

When diagnosing, separate the variables: test raw transfer speed with a large download through the proxy, test latency with small requests, and compare both against a direct connection to see which leg of the path is the real constraint.