What Is the Bandwidth-Delay Product?
The bandwidth-delay product (BDP) is a networking concept that describes the amount of data that can be "in flight" on a network path at any given instant. It is calculated by multiplying the link capacity (bandwidth) by the round-trip time (RTT):
BDP = Bandwidth (bits/second) x RTT (seconds)
The result is in bits. Divide by 8 to convert to bytes. For example, a 100 Mbps link with a 50 ms RTT has a BDP of 100,000,000 x 0.050 = 5,000,000 bits = 625,000 bytes (625 KB).
Why It Matters for TCP Performance
TCP uses a sliding window to control how much data can be sent before an acknowledgement (ACK) must be received. If the TCP receive window is smaller than the BDP, the sender will fill the window and then have to pause and wait for an ACK before sending more. This creates gaps in the data stream and means the link is never fully utilised.
The classic default TCP receive window is 65,535 bytes (64 KB), which was set when networks were slow and RTTs were short. On a modern gigabit link with 100 ms RTT, the BDP is 12.5 MB, so the 64 KB default allows the sender to keep only 0.5% of the pipe full. This is a major throughput bottleneck.
TCP Window Scaling (RFC 1323 / RFC 7323)
RFC 1323 (updated by RFC 7323) introduced TCP window scaling, which allows the receive window to be up to 1 GB. Most modern operating systems enable window scaling by default. However, some firewalls, routers, or older network equipment may strip the window scale option, causing performance problems on high-BDP paths. If large-file transfers are slow but ping latency is fine, a misconfigured or capped TCP window is a common cause.
Long Fat Networks (LFNs)
A network path with a large BDP is called a "long fat network" (LFN, sometimes pronounced "elephant"). This typically means either a very high bandwidth link, a very high latency path, or both. Satellite links are classic examples: a geostationary satellite link at 600 ms RTT and 10 Mbps has a BDP of 750 KB, meaning the sender must have that much data outstanding to keep the link busy. Dedicated transport protocols (such as SCP, QUIC, or BBR-based TCP) often outperform standard TCP on LFN paths.
Worked Example
Using the default inputs of 100 Mbps bandwidth and 50 ms RTT:
- Bandwidth in bits/s: 100,000,000 bits/s
- RTT in seconds: 0.050 s
- BDP = 100,000,000 x 0.050 = 5,000,000 bits
- BDP in bytes = 5,000,000 / 8 = 625,000 bytes
- BDP in KB = 625,000 / 1,000 = 625 KB
- BDP in MB = 625,000 / 1,000,000 = 0.625 MB
- The default 64 KB TCP window would limit throughput to: 65,535 x 8 / 0.050 = approx 10.49 Mbps, not 100 Mbps
To achieve full 100 Mbps throughput, the TCP window must be at least 625 KB, which requires window scaling to be enabled.
Related Calculators
Sources and method: RFC 7323 "TCP Extensions for High Performance" (tools.ietf.org/html/rfc7323). RFC 1323 "TCP Extensions for High Performance" (original, superseded). Mathis, Madhavi, Peterson, Sikdar (1997) "The Macroscopic Behavior of the TCP Congestion Avoidance Algorithm", ACM SIGCOMM CCR. Formula: BDP (bits) = Bandwidth (bits/s) x RTT (s); BDP (bytes) = BDP (bits) / 8.
This calculator provides theoretical BDP values based on the formula BDP = bandwidth x RTT. Actual achievable throughput depends on TCP implementation, window scaling support, congestion, packet loss, and network equipment. Use measured RTT values (from ping or mtr) for the most accurate results.
