Performance

Throughput Calculator

Calculate real network throughput accounting for protocol overhead. Understand efficiency losses from Ethernet, IP, and TCP/UDP headers.

Line Rate

MTU Size

Protocol Stack

Throughput Analysis

Effective Throughput
0.00
Mbps
Protocol Efficiency0.00%
Total Overhead
0
bytes/packet
Payload Size
0
bytes/packet
Packets/Second
0
at line rate
Bits/Packet
0
on wire
Overhead Breakdown
Ethernet38 bytes
IPv420 bytes
TCP20 bytes
Total0 bytes

Jumbo Frames Benefit

Standard (1500 MTU)~94.9% efficiency
Jumbo (9000 MTU)~99.1% efficiency

Jumbo frames reduce overhead ratio by transmitting more data per packet. Requires end-to-end support on all network devices.

Protocol Overhead Reference

Layer 2

Ethernet38B
802.1Q VLAN42B
QinQ (802.1ad)46B

Layer 3

IPv420B
IPv4 + Options60B
IPv640B

Layer 4

TCP20B
TCP + Timestamps32B
UDP8B

Throughput at Different Speeds

Link Speed Max PPS (64B) Max PPS (1518B) Goodput (TCP/IPv4)
100 Mbps 148,810 8,127 94.9 Mbps
1 Gbps 1,488,095 81,274 949 Mbps
10 Gbps 14,880,952 812,744 9.49 Gbps
100 Gbps 148,809,524 8,127,441 94.9 Gbps

PPS = Packets Per Second. 64-byte frames are worst case for routers/switches (highest PPS). Goodput assumes 1500 MTU with TCP/IPv4/Ethernet.

Bandwidth vs Throughput

  • Bandwidth: Maximum theoretical capacity of the link
  • Throughput: Actual data transfer rate achieved
  • Goodput: Application-level data rate (excludes all headers)
  • Wire rate: Actual bits on the physical medium including preamble

Improving Efficiency

  • Jumbo frames: Use 9000 MTU where supported for ~99% efficiency
  • TCP offload: Let NICs handle segmentation to reduce CPU overhead
  • Packet coalescing: Combine small packets where latency allows
  • UDP for bulk: 8-byte header vs 20+ for TCP when reliability not needed