Please try to explain to yourself on how a switch with 4x 10 GbE and 24x/48x 1 GbE ports should ever reach the 128 Gb/s resp. 172 Gb/s as indicated on the Smart Switch Series S3300 - S3300-28X, S3300-28X-PoE+, S3300-52X and S3300-52X-PoE+ Data Sheet.

 

A switch can not exceed the sum of bandwidth available on all ports, in full duplex multiply with two. With 24 GbE ports and four 10 GbE ports: 24x2 + 4x10x2 == 108 Gb/s, with 48 GbE and four 10 GbE ports == 128 Gb/s.

 

These S3300 models are probably built on switch cores allowing either more, or faster MII Signals, however these are not accessible and can't be used in realty therefore. Amazing this wrong data is kept carried forward in the dats sheet revision DS-S3300Series-14Jan21.

 

*===

The backplane bandwidth of the switch is the maximum amount of data that can be throughput between the switch interface processor or the interface card and bus. The backplane bandwidth indicates the total data exchange capacity of the switch, and the unit is Gb/s, also called the switching bandwidth. The backplane bandwidth of a general switch ranges from several Gb/s to hundreds of Gb/s. The higher the backplane bandwidth of an switch, the stronger the ability to process data, but the higher the design cost.

 

The utilization of backplane bandwidth resources is closely related to the internal structure of the switch. The internal structure of the switch mainly includes the following: First, the shared memory structure, which relies on the central switching engine to provide high-performance connections of all ports, and the core engine checks each input packet to determine the route. This method requires large memory bandwidth and high management costs, especially with the increase of switch ports, the price of the central memory will be high, so the cow of the switch becomes the bottleneck of performance realization; the second is the cross bus structure, which a direct point-to-point connection can be established between ports, which is very good for single-point transmission, but not suitable for multi-point transmission; the third is the hybrid crossover bus structure, which is a hybrid crossover bus implementation. The integrated cross-bus matrix is divided into small cross-matrixes, which are connected by a high-performance bus in the middle. The advantage is that the number of crossover buses is reduced, the cost is reduced, and the bus contention is reduced; but the bus connecting the crossover matrix becomes a new performance bottleneck.

 

When purchasing an switch, you can also calculate the backplane bandwidth through a formula based on the number of ports and port rates to determine whether the switch meets the requirements and whether it is a wire-speed switch. How to calculate it specifically?

 

1) Line-speed backplane bandwidth

 

Calculate the total network bandwidth available from all port numbers on the switch. The calculation method is the number of ports * relative port number speed * 2 (full duplex mode) If the total network bandwidth ≤ tolerance backplane bandwidth, then the backplane bandwidth is wire-speed.

 

2) Layer 2 packet forwarding wire speed

 

The second layer packet forwarding rate = the total number of Gigabit network card port numbers × 1.488 Mpps + the total number of 100 M port numbers * 0.1488 Mpps + the number of other types of ports * relative calculation method, if this speed can be less than or equal to the tolerance layer two packet forwarding speed, then the switch can guarantee the wire speed in the case of doing the second layer exchange.

 

3) Layer 3 packet forwarding wire speed

 

The third-layer packet forwarding rate = the total number of gigabit network card port numbers × 1.488 Mpps + the total number of 100-megabit port numbers * 0.1488Mpps + the number of other types of ports * relative calculation method, if this speed can be less than or equal to the tolerance of the three-layer packet forwarding speed, then the switch can guarantee the wire speed in the case of doing Layer 3 interchange.

 

So, how did the gigabit port packet forwarding rate of 1.488 Mpps get it?

 

The evaluation index of the packet forwarding line speed is based on the number of 64 byte data files pushed per unit time as the measurement standard. For Gigabit Ethernet, the calculation is as follows:

 

1000000000/8/(64+8+12)=1 488 095.2380952 pps is 1.488 Mpps, of which 1000000000: 100 Mbps port transmission rate, the first 8: 8 bits per byte; 64: the minimum length of the Ethernet frame; Two 8: Ethernet frame preamble, 12: frame gap

 

When the Ethernet frame is 64byte, consider 8byte frame header and 12 byte frame gap of the fixed overhead. So a wire-speed Gigabit Ethernet port forwarding 64byte packet when the packet forwarding rate of 1.488 Mpps. Fast Ethernet wire-speed port packet forwarding rate is exactly one-tenth of Gigabit Ethernet, which is 148.8 kpps.

 

• For 10 Gigabit Ethernet interfaces, the packet forwarding rate of a wire-speed port number is 14.88Mpps.
• For Gigabit Ethernet, the packet forwarding rate of a wire-speed port number is 1.488Mpps.
• For a Fast Ethernet interface, the packet forwarding rate of a wire-speed port number is 0.1488Mpps.

Therefore, if the above three standards can be considered, then we can say that the switch truly guarantees linearity and no blockage. Therefore, when purchasing an switch , it is necessary to select a switch that can meet the requirements of the backplane bandwidth and packet forwarding rate according to the number of ports and port rate of the switch, otherwise it will easily form a bottleneck in the system.

===

*borrowed and adopted from medium.com here.