Green development of LAN switches is the way out

We all know that there are many reasons that affect the power consumption of LAN switches. For common users, we try our best to reduce the energy consumption. Many network equipment manufacturers today promise that their networks and switching devices are energy-saving, but few vendors can prove their actual use. Therefore, enterprise users must test the energy efficiency of LAN switches through a certain performance test. This test is basically the same as the test method for determining the speed and function. When talking about lan energy consumption efficiency and time consumption, we can consider the following measurement methods:

1. system throughput

In general, throughput is the most important test factor in the performance test of LAN switches. This is also an important factor in measuring power consumption. These test numbers are combined with the measured power consumption to calculate the throughput per watt of electricity. We recommend that the tester calculate the bandwidth (Mbps/W) forwarded per watt of power per second ). Of course, you can also choose Gbps. You can also use raw data to calculate the amount of money required to transmit data per shard per second.

Another important reason for throughput measurement is that some manufacturers choose to implement a network connection structure that does not provide wire speed throughput to all additional interfaces. These LAN switches have many plug-in modules or backboards, which may use less power than LAN switches that provide full line rate throughput. Without throughput measurement, we mistakenly conclude that low-capacity devices are more efficient than switches capable of delivering larger throughput when transmitting data from the same number of ports. Only by taking the throughput calculation into account can we associate the reached throughput with the consumed power.

2. Power Factor

In this case, it is equally important to determine the efficiency of the electric energy consumed by the device under test as that consumed by the measurement. Therefore, when testing the energy efficiency of LAN switches, consider the "power factor ". According to Wikipedia, power factor refers to the actual power of the "apparent power" load. The underutilized power means that a device consumes more energy than it actually can. In the long term, the cost is much higher than the required cost.

The power factor is a number between 0 and 1, where 1 represents the maximum or 100% efficiency. Some testing tools automatically calculate this value. The most obvious power consumed by a system is the product of the RMS Value of the voltage and the current flowing through the device. Here we assume that the waveform is of the same phase. The power supplier uses this value to evaluate the total power consumption. The problem is that, due to a series of complex devices in the network, the voltage and current waveforms are not in the same phase. This measurement is only available when the AC power supply is involved, and this method is not available for DC systems.

3. Communication Load

Communication load must be considered when testing power consumption. It is important to consider different load levels to obtain precise information about the energy consumption of different levels of network activity. Note that the connected port and open port State not only mean that the cable is connected, but also indicate that the physical layer and the MAC layer are active.

1. None connected

2. None of the active (connected and open)

3 activity 0%

4 activity 50%

5 activity 100%

4. Frame/packet size

This is critical. Historically, two-layer and three-layer switches are tested through a series of frames/data packets of different sizes. The minimum valid frame is 64 bytes, and the maximum standard frame is 1518 bytes, in addition, the size of some tested data frames has also changed. The most commonly used data frames include 128, 256, 512, and 1024 bytes. Some tests also include non-standard large data frames. For example, some tests use a test frame of up to 16000 bytes. However, a typical test uses a data frame of 9 KB or 9128 bytes. Of course, if we only want to test the power consumption of Layer 2 or Layer 3 switches, we do not need to use so many data frames for comprehensive testing.

Although there is no industry standard for testing the power consumption of large data frames, users should remember that in general, we should avoid testing with only 64-byte data frames. This test forces the switch to process the maximum number of data packets (the number of data packets is naturally larger), so that the power consumption reaches the maximum. In contrast to the above situation, if your test uses a data frame of 1518 bytes or larger, this reduces the number of data frames to be processed per unit of time (of course, this depends on the device architecture used), thus reducing power consumption. In any case, the tester should pay attention to the size of the data frame used in a specific test.

Tests on layer-4 and higher-level switches need to take into account the actual communication flow (for example, session establishment, data transmission, and session removal ). In terms of its nature, this communication mainly consists of multiple data frames (packets. Therefore, if the tester wants to test such a high-rise switch, it is not advisable to consider only one type of data frame. However, we need to realize that the processing of smaller and larger data units does exist in layer-4 to layer-7 switch tests. This type of data unit is called "object size" in the test, that is, the size of the object that the switch returns to the client from the server. It is very important that these objects are generally larger than the maximum data frame (1518 bytes) over standard Ethernet.

5. Communication types considering the Energy Efficiency Test of LAN switches

Depending on the communication properties, the switch must process Communication in hardware or software or in combination of software and hardware. Although the manufacturer of a vswitch rarely discloses its details, it is important for testers to understand that software and hardware can have a significant impact on the energy consumption of the vswitch. When a switch cannot process a communication that relies on a hardware chip, it must rely on software running on the main processor, which must increase the CPU burden and increase energy consumption.

Therefore, it is very important to measure the time consumption, and the communication type and the communication combination type must be suitable for your purpose. Although most LAN switches Process Layer-2 Communication in hardware, some switches process some or all of the layer-3 functions in the main processor. Sometimes it is not easy or even confusing to understand which layer of communication is being processed. The key is to remember that not the content decides the layer, but the switch function and settings determine the layer. For example, we can pass the layer-7 http Communication through the layer-2 switch, but this switch can only make a decision based on the layer-2 information. Therefore, the results are the same, regardless of whether the communication contains application information or the second-level address information.

Remember, because you want to transmit upper-layer communication, you cannot simply assume that the switch is processing data at a specific level, unless you have developed a test plan, it also proves that you have guided the communication according to the specific level of content in the protocol stack. For example, we can design many Layer 7 tests to instruct the switch to send communication to a specific port based on the requested web page. In this case, by verifying whether the server has received a "get" request, the tester can easily prove whether the process is being executed at a certain level. Next we will first explain the measurement indicators and other considerations of the power consumption and efficiency of LAN switches. The following describes how to measure the power consumption of LAN switches. What factors need to be considered when testing the power consumption of LAN switches?

There are many factors that can affect the way LAN switches consume power. The network interface is either copper or optical fiber media, which directly affects the power consumption, as well as the number of active ports and the number of modules that need to transmit software on some layers of the network. In addition, the deeper the communication check, the more power consumption. Layer-4 or above switches need to be tested. These LAN switches make decisions based on the height of data in the protocol stack and the depth of data packets. This leads to a different data encoding path, which affects power consumption. LAN Switch Power Consumption Test indicators:

1. Steady-State Power Consumption

Steady-state power consumption is often used in performance tests. However, from the perspective of power consumption, this is inaccurate. For example, a running fan will undoubtedly increase power consumption. Some manufacturers may allow the fan to run when the system is started, thus "testing" the fan (even if no heat dissipation is required ). Users can test power consumption five minutes after the device is powered on, which temporarily increases power consumption.

In addition, we also need to determine whether the steady-state power consumption is based on a device with an idle port or a device that is processing data communication. Regardless of the definition, it is important to note that cost calculation based on steady-state power consumption alone cannot be very accurate, because over time, the LAN switch does not maintain the status of a single communication load.

2. modules/interfaces

To obtain accurate energy consumption figures, the LAN switch test must note all modules in use in the system. Even LAN switches with stackable switches or fixed ports may only use certain modules, which may potentially affect power consumption. In general, this option includes the uplink port and the stacked port used to connect a vswitch baseboard to another vswitch baseboard.

3. Influence of Power Supply

For a system that provides multiple power supplies, when the load is between 50% and 90%, the power supply is generally the most effective, so the power supply option is very important. For a modular system with low load, selecting the highest power function will lead to inefficient use of power supply and higher power consumption.