Ethernet line rate refers to the maximum nominal speed at which an Ethernet connection can transmit data. For example, a 100 Gigabit Ethernet (100GbE) connection has a line rate of 100 gigabits per second (Gbps), while a 400GbE connection supports 400 Gbps.
However, the speed of an Ethernet connection does not necessarily represent the amount of useful data an application can transfer or the amount of traffic a network device can process.
Understanding the difference between line rate, throughput and packet-processing performance is important when evaluating routers, switches, network interface cards (NICs) and other networking equipment.
What Does Line Rate Mean in Ethernet?
In Ethernet networking, line rate describes the nominal maximum data transmission rate of an Ethernet interface.
Common Ethernet interface speeds include:
| Ethernet standard | Nominal line rate |
|---|---|
| 1 Gigabit Ethernet (1GbE) | 1 Gbps |
| 10 Gigabit Ethernet (10GbE) | 10 Gbps |
| 25 Gigabit Ethernet (25GbE) | 25 Gbps |
| 100 Gigabit Ethernet (100GbE) | 100 Gbps |
| 200 Gigabit Ethernet (200GbE) | 200 Gbps |
| 400 Gigabit Ethernet (400GbE) | 400 Gbps |
For example, a router equipped with a 100GbE interface can transmit or receive Ethernet traffic at a nominal rate of 100 Gbps.
However, that does not automatically mean the router can forward, inspect or filter all incoming traffic at that speed.
The device must also have enough processing capacity to handle the packets it receives.
This is why line rate is used not only to describe the speed of a network connection but also to evaluate the performance of networking equipment.
When a router or switch is described as line-rate capable, it generally means the device can process traffic at the full supported interface speed under specified conditions.
Ethernet Line Rate vs Bandwidth vs Throughput
Although these terms are closely related, they describe different aspects of network performance.
Line rate refers to the nominal maximum transmission speed of an Ethernet connection.
Bandwidth, when discussing Ethernet links, commonly refers to the available transmission capacity, expressed in bits per second.
Throughput refers to the amount of data successfully transferred or processed over a period of time.
For example, imagine two servers connected through a 100GbE network.
The Ethernet interfaces support a nominal line rate of 100 Gbps in each direction, but an application transferring files between those servers may achieve less than 100 Gbps of useful data throughput.
Several factors can contribute to this difference, including Ethernet framing overhead, higher-layer protocol headers, network congestion and the processing capabilities of the connected devices.
A network connection may therefore support 100 Gbps even when the useful application throughput is lower.
This distinction becomes particularly important when assessing high-speed network equipment.
Why Is Actual Throughput Lower Than Ethernet Line Rate?
Ethernet does not transmit application data alone. Each Ethernet frame includes additional information required to deliver traffic across the network.
For example, a standard Ethernet frame includes source and destination MAC addresses, a frame type or length field, and a frame check sequence (FCS).
Ethernet transmission also requires a preamble and inter-frame gap.
These elements consume part of the available connection capacity.
As a result, not every transmitted bit represents useful application data.
A simple example illustrates this difference.
A minimum-sized Ethernet frame is 64 bytes long, including its FCS. When the standard 8-byte preamble and 12-byte inter-frame gap are included, each frame occupies 84 bytes of transmission capacity.
Only part of that transmission capacity carries the data being transported.
Larger frames generally use the available bandwidth more efficiently because the fixed overhead represents a smaller proportion of each frame.
Higher-layer protocols such as IP and TCP also introduce additional headers, reducing the proportion of bandwidth available for application payloads.
Other factors, such as packet loss, congestion and endpoint processing limitations, can further reduce observed throughput.
Is Wire Speed the Same as Line Rate?
In networking discussions, wire speed and line rate are often used interchangeably.
Both terms commonly describe a network device's ability to transmit or process traffic at the full supported speed of its interfaces.
For example, a switch advertised as providing wire-speed forwarding is generally expected to forward traffic at its rated interface capacity under the vendor's stated conditions.
However, achieving wire-speed performance depends on the device's hardware architecture and the operations performed on each packet.
A switch may forward packets at line rate under one set of conditions but achieve lower throughput when more demanding processing features are enabled.
For this reason, line-rate and wire-speed claims should always be considered alongside the traffic characteristics and test conditions used to measure them.
How Packet Size Affects Ethernet Performance
An important consideration in high-speed networking is that the same Ethernet line rate can represent very different packet-processing workloads.
A network carrying large packets processes fewer packets per second than a network carrying the same amount of traffic using small packets.
For example, a 100GbE connection carrying minimum-sized Ethernet frames can transmit approximately 148.8 million packets per second (Mpps).
If the same connection carries larger frames, the number of packets transmitted per second is much lower, even when the link is operating at full capacity.
The following table shows the theoretical packet rates for minimum-sized Ethernet frames.
| Ethernet line rate | Maximum packet rate (64-byte frames) |
|---|---|
| 10 Gbps | 14.88 Mpps |
| 100 Gbps | 148.81 Mpps |
| 200 Gbps | 297.62 Mpps |
| 400 Gbps | 595.24 Mpps |
These calculations include the standard Ethernet preamble and inter-frame gap, assuming continuous transmission of 64-byte Ethernet frames.
Packet size matters because network devices perform processing operations on individual packets.
Depending on the device, these operations may include reading packet headers, making forwarding decisions, applying filtering rules or updating counters.
A device that handles 100 Gbps of larger packets may not necessarily be capable of processing 148.8 million minimum-sized packets every second.
This is why network performance is often measured using both bandwidth (Gbps) and packet-processing rate (Mpps).
For a practical example of how engineers measure these limits, see Denys Haryachyy's article, Learning TRex: Generating 100GbE Line-Rate Traffic on a Mellanox ConnectX-5.
Can Network Equipment Really Operate at Line Rate?
Yes. Many modern routers, switches and network processing systems are designed to operate at line rate.
However, the ability to achieve full interface throughput depends on the underlying technology and processing workload.
Network switches and routers commonly use specialised hardware, including Application-Specific Integrated Circuits (ASICs), to forward packets efficiently.
Software-based packet-processing systems can also achieve high throughput using optimised software frameworks, suitable network interface cards and sufficient CPU resources.
For example, FD.io VPP (Vector Packet Processing) uses techniques such as processing packets in batches to reduce per-packet overhead.
Regardless of the architecture, advertised interface capacity should not be confused with verified processing throughput.
When evaluating networking equipment, engineers typically consider factors such as:
- Maximum throughput in Gbps
- Packet-processing capacity in Mpps
- Performance with different packet sizes
- Packet loss under high traffic loads
- Performance when additional processing features are enabled
Testing with different packet sizes is particularly important when evaluating devices that must handle demanding network workloads.
For a deeper technical explanation, Denys Haryachyy explores the practical challenges of achieving full 100GbE performance in Tuning a VPP Data Plane to 100GbE Line Rate: Cycle Budget, NIC Rings, and RSS.
Where Is Line-Rate Ethernet Performance Important?
Line-rate performance is especially relevant in networks where large volumes of traffic must be forwarded or processed without introducing bottlenecks.
For example, ISPs and data centres operate high-capacity Ethernet connections that carry traffic for many customers simultaneously.
Network equipment must be capable of keeping up with those traffic volumes while performing its required functions.
One application where this becomes particularly important is protection against Distributed Denial of Service (DDoS) attacks.
During a DDoS attack, a network may receive extremely high volumes of traffic or millions of small packets per second. A filtering system must be able to process this traffic efficiently, blocking malicious packets without becoming a bottleneck.
This is where inline DDoS filtering comes into play. Inline filtering systems process traffic directly in the network path, applying filtering rules and forwarding permitted packets. Their ability to operate at line rate is especially important in high-capacity networks.
To understand how this works in practice, read our guide to Line-Rate DDoS Filtering Explained: How High-Speed Packet Filtering Works.
FastNetMon is also developing high-performance inline packet filtering technology through FastACL Community, an open-source filtering engine built on FD.io VPP.
FastACL's published performance testing demonstrates how software-based packet processing can achieve high throughput on commodity hardware, while also highlighting why packet size and processing capacity are important performance considerations.
For readers interested in the technical implementation, Denys Haryachyy's article, Learning VPP: Filtering Packets at 100GbE Line Rate, explores how packet classification and filtering can be optimised for high-speed Ethernet traffic.
Conclusion
Ethernet line rate describes the nominal maximum transmission speed of a network connection, such as 10 Gbps, 100 Gbps or 400 Gbps.
However, a network interface's advertised speed does not automatically guarantee equivalent application throughput or packet-processing performance.
Ethernet and protocol overhead reduce useful data throughput, while packet size and processing complexity influence how many packets a networking device can handle.
Understanding these differences helps network engineers evaluate equipment performance, identify potential bottlenecks and assess whether a device can meet the demands of high-capacity networks.
For applications such as DDoS filtering, where processing large numbers of packets is essential, both line-rate throughput and packet-processing capacity are important considerations.
