How to measure the network delay in an Ethernet switch?

As can be seen from the previous chapter, switch delay is one of the key factors leading to network delay. So how do we measure switch latency?

Switch delay is measured from port to port on the Ethernet switch, and can be reported in multiple ways according to the forwarding method adopted by the switch. There are many ways to measure the network delay of an Ethernet switch, such as IEEE standard RFC2544, Netperf or Ping Pong. The IEEE standard RFC2544 provides an industry-approved method for measuring the delay of storage and forwarding equipment. Netperf can use request or response test (TCP_RR and UDP_RR) delay. Ping Pong is a method for measuring latency in high-performance computing clusters. This method measures the round-trip time of remote procedure calls (RPC) sent through the message passing interface (MPI).

How to reduce the network delay caused by the Ethernet switch?

In order to reduce the network delay of the Ethernet switch, the following methods can be used:

1.Expand network capacity
Expanding the capacity of an Ethernet switch is critical to reducing network latency. You can check your switch to see if it has the function of expanding network capacity. First, you need a fast engine. An Ethernet switch with zero packet loss will help the network achieve better performance. LACP is a standard protocol that can help establish better network performance through a trunk port. Fast (FS) S3900 series switches support LACP to increase bandwidth to improve network performance.

2.Use VLAN to assign network

Because the traditional flat network architecture can easily overload the switch link, the Ethernet switch with VLAN function can send traffic to an easily accessible location. There are many Layer 2 and Layer 3 Ethernet switches that have VLAN functions, which can segment traffic based on ports, dynamic VLAN assignment, protocols, MAC addresses, and other types.

3.Adopt cut-through forwarding technology

Cut-through forwarding technology is a method used for data packet switching to minimize network latency. Switches using cut-through forwarding technology have a fixed delay. Because the straight-through switch does not care about the overall size of the data packet, but only determines the forwarding direction based on the destination address. Therefore, its delay is fixed. Generally, after processing the destination address, the Ethernet switch will start forwarding the data packet before receiving the entire data packet, thereby reducing the waiting time for passing through the switch.Note, however, that it will not work when sending traffic from a slow port to a faster port or from one port to another port at the same speed.

The above method can minimize the network delay of the Ethernet switch. There are many low-latency Ethernet switches on the market, which help to obtain better network performance. However, in order to minimize the network waiting time, fundamentally speaking, not only must focus on the switches that make up the network, but also the waiting time of the system and changes in the waiting time must always be paid attention to.

To sum up
In general, the smaller the network delay, the better the network quality. At the same time, the delay of the network cannot be completely eliminated, but it can be minimized. An Ethernet switch with low latency and high bandwidth is a good choice, which helps minimize network latency.

JHA-MIG024W4-1U Managed Industrial Ethernet Switch