Switch) performance directly affects the entire network. Therefore, it is not only necessary to understand the Switch parameters, but also helps you make better choices that meet your actual needs. Currently, the price of ATM switches remains high and is mainly used in backbone networks or wide area networks of large networks. Therefore, we will focus on Parameter Analysis of Ethernet switches.
Selection of Backboard bandwidth and port Rate
The port speed of a vswitch has increased from 10 M and 1000 M to M, and the concept of a 10-Gigabit vswitch has been proposed. From the current hot spots of network applications, the 10G switch has been out of the market. In addition, due to the sharp reduction in the price of 10/100 M Adaptive Nic, users can enjoy fast Ethernet technology on the desktop, and more users will use Gigabit Ethernet Switching Technology on the backbone. M switches are generally used in backbone networks of large networks. Like ATM, they provide users with high-speed backbone bandwidth. The m switch will play a role in the backbone of Small and Medium-sized networks, or play the role of a secondary switch in large networks.
For a m switch, a common parameter is automatic rate adaptation, that is, the port rate of the switch can match the NIC and decide whether to connect at a rate of 10 m or M.
Although the port speed is very important, the factors that affect the switching speed apart from the number of packets per second on the port throughput, there are also: 1) Board bandwidth: Of course, the wider the better, it provides high-speed switching for your switch under high load; 2) packet forwarding rate: The number of packets forwarded by the switch per second.
We know that a Switch attaches each port to a backend Bus Core Bus with a high bandwidth. the backend Bus is connected to the Switch Engining engine, the encapsulated data packets from the port enter the switching engine through the backplane bus. Vswitches exchange data packets in three ways:
Cut through: After encapsulated data packets enter the switching engine, they are dropped to the backplane bus within the specified time and then sent to the destination port. This switching mode is fast, but is prone to packet loss;
Store & Forward: a buffer exists after the encapsulated data packet enters the switching engine and is forwarded to the backplane bus by the switching engine. This switching method overcomes packet loss but reduces the switching speed;
Fragment Free: a solution between the two.
Of course, not all vswitches support the above three switching modes. Some vswitches only support the first two switching modes and do not support Fragment Free.
Modular or fixed configuration?
Currently, there are only two types of switches on the market: one is a chassis switch, also known as a modular switch, and the other is a standalone fixed configuration switch.
The biggest feature of a chassis switch is its strong scalability. It provides a series of expansion modules, such as the Gigabit Ethernet module, FDDI module, ATM module, Fast Ethernet module, and card ring module, therefore, networks with different protocols and different topologies can be connected. Its biggest disadvantage is its high price. A chassis switch is generally used as a backbone switch.
Fixed configuration switches generally have fixed port configurations, such as Cisco's Catalyst 1900/2900 switch, 3Com's SuperStack II series, and Bay's BayStack 350/450 switch. The scalability of fixed configuration switches is obviously inferior to that of chassis switches, but the cost is much lower.
Therefore, when selecting a vswitch, you should consider whether to purchase a chassis or a fixed configuration vswitch based on your needs and expenses. In general, the expansion and redundancy of the central switch of a large network should be considered, and the chassis switch is suitable. The secondary switch or the backbone of a small or medium-sized network can adopt a simple and clear independent switch.
Dedicated chip and general Chip
A vswitch is actually a computer, so it also has its own processor CPU ). In a 100 M/M switch, the processor task is very heavy. Some switch manufacturers use general-purpose CPU chips. Because the general-purpose CPU chip is not designed for the switch, the work efficiency is relatively low. If multiple ports work at the same time, packet Loss and congestion may occur.
Most switch manufacturers use their own dedicated ASIC chips. This type of chip is designed for vswitches and is highly efficient. Therefore, when selecting a vswitch, pay special attention to whether the switch uses an ASIC chip.
Which VLAN do you need?
To reduce collision and broadcast storms and enhance security, users usually require the switch to Have VLAN division function, that is, to achieve subnet division on the switch. A VLAN is a port on a vswitch that can exchange a single broadcast packet. When a packet is broadcast from a port of a VLAN, the switch receives the packet and copies it to all ports included in the VLAN.
Generally, different VLANs cannot communicate with each other, but some switches support a port that can belong to either the VLAN or another VLAN.
Some switches depend on the attached protocols used for VLAN information communication between switches, allowing a VLAN to span over the ports of multiple switches.
VLAN division methods are usually as follows: the earliest VLAN division is Based on Port Based), that is, VLAN Division through ports; the current switch also supports MAC address Based) and IP address Protocol Based) to divide VLANs. Some newer switches can also manage VLANs through Policy Service (Policy Servie), further simplifying VLAN division and management.
Most switch VLANs follow the IEEE 802.1Q standard, while some switches follow the CGMP Cisco Group Management Protocol proprietary standard.
Layer-3 switching or layer-4 switching?
A user divides a subnet VLAN on a L2 Switch. The communication between subnets depends on the communication between routers. This is the practice of a traditional network. The low efficiency and high latency of routers make it the bottleneck of network traffic. As a result, a new device with Layer 2 vswitch and Layer 3 vro features, and Its latency is much lower than that of traditional vrouters. This is the third-tier switch. Layer-3 switches are suitable for interconnection between subnets with multiple subnets, such as backbone switches of large enterprise networks or campus networks.
It is worth mentioning that there are no accepted standards for the implementation of layer-3 Exchange Technology. Different vendors have different practices, such: ATM Forum MPOA, 3Com Fast IP, Cisco Tag Switch, Ascend IP Navigator, and so on, of course, there is also a great development potential of MPLS.
Some layer-3 switches have the ability to identify layer-4 protocol ports, which are called layer-4 switches. Basically, the layer-4 Exchange is actually a layer-3 exchange, but there are some value-added software. The layer-4 switch does not actually work on the transport layer. It performs the switch operation on the layer-3, but is more sensitive to the layer-3 switch.
Connections between vswitches
Do vswitches need to be connected to each other? Some people think that connecting them with a pair of crossover twisted pair wires 1, 2, 3, and 6 is all right. Otherwise, redundancy and bandwidth bottlenecks must be considered for the connection between vswitches.
Redundant connections break through the bottleneck
We know that loops are not allowed in the Ethernet environment, and the generation Tree Spanning Tree can be used to achieve redundant connections between switches and avoid loops. Of course, this requires the switch to support the Spanning Tree.
However, the Spanning Tree redundant connection works in the Stand By state, that is, except for one link, the other links are actually in the standby Stand By state, which obviously affects the transmission efficiency. Some of the latest technologies, such as FECFast Ethernet Channel), ALBAdvanced Load Balancing) and Port Trunking technology, allow each redundant connection link to achieve Load Balancing. The FEC and ALB technologies are used to connect a vswitch to a Server Switch), while the Port Trunking technology is used to connect a vswitch to a Switch ). Through the redundant Port Trunking connection, vswitches can achieve connections that are several times faster than the line rate bandwidth.
Switches that provide stack interfaces can be connected through dedicated stack lines. Generally, the stacked bandwidth is dozens of times the speed of the switch port. For example, for a m switch, the bandwidth between the two switches can reach hundreds of megabytes or even Gbit/s.
The stacking of multiple switches is achieved by connecting a multi-port stacked parent module that provides backplane bus bandwidth with a single-port stacked sub-module, and inserting different switches to stack switches. The uplink switch can connect to the backbone switch through the uplink port. For example, a vswitch with 24 10 m ports and 1 m ports can connect to the master vswitch with M ports.