How to Choose an Etherent Switch?
When choosing the models of Ethernet switch, a lot of technical parameters should be taken into account, which includes 100/1000/10000 ports, electrical ports/optical ports/PoE ports, the number of ports, MAC address table depth, forwarding latency, cache size, VLANs, isolation, etc.
Usually, the weak current demand of a building or a project often covers more than one weak current system. Taking a logistics park as an example, common systems in logistics parks include building automation systems, intelligent lighting systems, integrated wiring systems, video surveillance systems, and so on. In a video surveillance system, we need to know the installation location and quantity of each camera, calculate the stream size of the cameras, and select the switch based on these data.
Should you choose Gigabit Switch or 100M Switch?
In the network of the video surveillance system, a large amount of continuous video data needs to be transmitted, which requires the switch to have the ability to transmit data stably. The more cameras connected to the switch, the greater the amount of data flowing through the switch. Think of the switch as a pool, and the code flow is the water in the water pipe leading to each pool. Once the pool is full, the excess water will overflow out of the pool and be lost. In the same way, if the amount of data forwarded by the camera under the switch exceeds the forwarding capability of a certain port, it will also cause this port to discard a large amount of data. For instance, if a 100M switch forwards more than 100M of data, it will cause a large amount of packet loss, resulting in a blurry screen.
How to choose Core Layer Switch?
Thus, under what kind of circumstances should you choose a Gigabit switch?
The answer depends on the size of the data forwarded by the camera’s uplink port. If the forwarded data volume of the uplink port is larger than 70M, you should choose a Gigabit switch or a Gigabit uplink switch.
The code stream of a single camera = main code stream + sub-code stream. The main code stream is generally a high-definition picture, which is used for video recording and real-time preview of the picture; the sub-code stream is generally a standard definition picture display, and used for network transmission or video surveillance screen display. It's usually 0.5M. Now the mainstream encoding technology of the camera is mainly H.265. The following is the reference of commonly used camera code streams:
40 H.265 200W pixel cameras (2+0.5) are connected to one switch, and the forwarding rate of the uplink port is 2.5*40=100M>70M, so a Gigabit switch is required. If one port of the switch needs to be Gigabit, then you need to select a full Gigabit switch, or a switch whose uplink port must be a Gigabit port.
Then, does it comes to the final when selecting the access layer switch according to the above calculation?
No. For a large and medium-sized network monitoring system in a district, in addition to the access layer switches as "limbs", a "brain" is also needed to control them. The system will be flattened according to the " access-aggregation-core ". The core switch is the "brain", which is the forwarding center of the entire network video traffic and bears a large amount of data. Therefore, it is necessary to ensure that each port of the switch meets the needs without bottlenecks.
Combined with the above calculation method, it is easy for everyone to have some misunderstandings about the selection of core switches. For example, if there are 300/500 cameras, according to the calculation method of 500*2.5M=1250M, the forwarding speed of the Gigabit port is much higher than that of the Gigabit port, and the Gigabit port cannot be used. So is a 10 Gigabit port needed?
Gigabit ports are not necessarily required. In fact, in many large-scale network video surveillance systems, video traffic must be distributed on multiple ports and forwarded by multiple ports. Any two Gigabit ports of a full Gigabit switch can achieve 1000M bidirectional transmission. The total throughput (full load) is generally less than or equal to the backplane of the switch bandwidth.
Therefore, when selecting a core switch, according to the number of IPCs, the following suggestions are recommended:
①100~200 units, Gigabit managed switches are recommended
② 200~500 sets, three-layer managed switches are recommended
At present, the Layer 2/3 managed full Gigabit switch is suitable for the core exchange of the monitoring network and undertakes large-capacity data exchange to form various networks. For a large or very large (300~1000) monitoring network, it is necessary to use a three-layer switch to divide the network segment and it is recommended to use a three-layer switch.