How to use asymmetric switches to improve network performance

Asymmetric switches are still commonly used, so I have studied how to use asymmetric switches to improve network performance. Here I will share with you, hoping to help you. Vswitches can be divided into two types based on the speed of each port of a vswitch. Symmetric switches and non-symmetric switches. Symmetric switches provide exchange connections between ports with the same bandwidth. For example, all ports are 100 Mb/s ports. Asymmetric switches have different bandwidths on different ports. For example, some ports have 100 Mb/s while others have 10 Mb/s.

For enterprises with tight pockets, using asymmetric switches to improve network performance is a good choice. For example, asymmetric switches can be used to connect servers to clients. For example, when I used to work as a network administrator in a small business, the switch was still a relatively expensive device. An enterprise has a file server. In order to improve the efficiency of the file server, but due to financial restrictions, I prefer to use asymmetric switches to Improve the access efficiency of the file server. Connect the high-speed port to the file server, and connect the low-speed port to the client. When multiple employees access the file server at the same time, more bandwidth is required to be allocated to the switch port connected to the server to prevent traffic bottlenecks on the port. In this case, the port connected to the file server has enough bandwidth to accommodate users' access requests, thus improving the efficiency of employees' access to the file server.

How an asymmetric switch works

Ethernet switches generally use the buffer technology to store and send data packets to the appropriate port or multiple ports. This space for temporary data storage is called the memory buffer zone. The memory buffer is generally used to forward data packets, Port-based storage buffer period and shared storage buffer. Assume that A vswitch has only three interfaces: A, B, and C. Now let's assume that there is data sent from Port A of the switch to port C. How does this storage buffer work?

If the vswitch uses a port-based storage buffer, data packets are stored in a queue connected to a specific port. That is to say, when the data packet enters from Port A of the switch and goes out of port C, the data is first stored in the memory buffer zone of port, instead of being directly forwarded to the memory buffer in the outbound port C. The switch must first determine whether another packet exists in the memory buffer of port. According to the principle of first-in-first-out. After all the preceding data is sent, the data packet will be sent out on port C. Therefore, this may lead to data delay. When A port C or A is busy, this delay will be serious. In addition, the size of the memory buffer is generally limited by the port. In this case, if the data is sent from the 100 Mbit/S port to the 10 Mbit/S port, the data packet loss will be serious. Therefore, Port-based storage buffers are generally used on symmetric switches instead of asymmetric switches.

Asymmetric switches are generally used in shared-based storage buffers. Shared storage buffer refers to a dedicated area on a vswitch to temporarily store these packets. This area is shared, and all ports of the vswitch can be accessed. This port-based storage buffer is essentially different. In the latter case, each storage buffer is independent, and ports cannot access each other, but only ports actively send data packets.

The other difference is that the capacity of port-based storage buffers is fixed, while the storage capacity of port-based storage buffers varies according to the port requirements during the shared port buffer period, and dynamic allocation. For example, if a 100 Mbit/S port of the vswitch needs to send data to the 10 Mbit/S port, the shared storage buffer will allocate enough storage capacity to it, let it put all the data packets in one time, and then wait in the shared storage buffer, and send them out through the port of 10 Mbit/S. The advantage of doing so is that data packet loss can be greatly reduced. This is very useful for the normal operation of asymmetric switches, so that packets in the 100 Mbit/S speed port can be successfully sent to the 10 Mbit/S port.

Understanding the working principle of asymmetric switches has two main functions:

First, we use asymmetric switches to identify authenticity. When I purchased an asymmetric switch, I found that some non-symmetric switches with fake cards were mixed with fish. Although the port identification or product manual supports asymmetric switching, the Port-based storage buffer technology is actually used. If such an asymmetric switch is applied to an enterprise network, it will not only not improve the network performance of the enterprise, but also cause frequent data packet loss, it causes more problems for network communication. Therefore, when purchasing an asymmetric switch, the network administrator should confirm whether the purchased asymmetric switch actually adopts the shared storage buffer technology.

Second, it also serves as a reference for our network deployment. To some extent, asymmetric switches can improve the access efficiency between the server and the client. However, they also have a degree that will not be infinitely improved. How many enterprise network application performances can be improved after an asymmetric switch is used depends on the storage buffer capacity and port speed of the asymmetric switch. With this understanding, our network management personnel can purchase asymmetric switches based on the actual situation of the enterprise.