Link aggregation, redundancy, stacking, and hot backup of core switches

1. Link Aggregation

Link aggregation is the combination of two or more data channels into a single channel that appears as a single higher bandwidth logical link.

Link aggregation is generally used to connect one or more devices with large bandwidth requirements, such as servers or server clusters connected to a backbone network. It can be used to expand link bandwidth and provide higher connection reliability.

1. Example

The company has two floors, each running a different business. Originally, the networks of the two floors were separate, but as they are all the same company, they will inevitably have business dealings. At this time, we can connect the networks between the two floors to enable high-speed communication between the interconnected departments.

As shown below:

As shown in the figure above, SwitchA and SwitchB are connected to VLAN10 and VLAN20 respectively through Ethernet links, and there is a large amount of data traffic between SwitchA and SwitchB.

The user hopes that SwitchA and SwitchB can provide a larger link bandwidth to enable communication between the same VLANs. At the same time, the user also hopes that a certain degree of redundancy can be provided to ensure data transmission and link reliability.

Create an EtherChannel interface and add member interfaces to increase the link bandwidth. Configure EtherChannel1 on the two switches respectively. Add the ports of the three lines that need to communicate to EtherChannel1 respectively, set the port trunk, and allow the corresponding VLAN to pass; in this way, the networks of the two buildings can communicate normally.

2. Implement configuration steps:

Create EtherChannel1 on SwitchA and configure it in LACP mode. The configuration process of SwitchB is similar to that of SwitchA and will not be repeated here.

Switch A

 switch > enable
 switch# conf t
 switch(config)# hostname SwitchA
 SwitchA(config)# interface range g0/0/1-3
 SwitchA(config-if-range)# channel-group 1 mode active
 SwitchA(config-if-range)# interface Port-channel1
 SwitchA(config-if)# switchport trunk encapsulation dot1q
 SwitchA(config-if)# switchport mode trunk

SwitchB

 switch > enable
 switch# conf t
 switch(config)# hostname SwitchB
 SwitchB(config)# interface range g0/0/1-3
 SwitchB(config-if-range)# channel-group 1 mode active
 SwitchB(config-if-range)# interface Port-channel1
 SwitchB(config-if)# switchport trunk encapsulation dot1q
 SwitchB(config-if)# switchport mode trunk

2. Link Redundancy

In order to maintain the stability of the network, in a network environment composed of multiple switches, some backup connections are usually used to improve the efficiency and stability of the network. The backup connection here is also called a backup link or redundant link.

3. Switch stacking

Multiple switches can be stacked into one logical switch by connecting them with proprietary stacking cables.

All switches in the logical switch share the same configuration and routing information. Adding or removing individual switches to the logical switch does not affect its performance.

The stacked switches are connected via two loops. The switch hardware is responsible for load balancing data packets on the dual loops.

The loop here acts as the backplane of this large logical switch. When both loops are working properly, the data packet transmission rate on this logical switch is 32Gbps.

When a data frame needs to be transmitted, the switch's software calculates which ring is more available, and then the data frame is sent to that ring.

If a stacking cable fails, the switches at both ends will detect the failure and disconnect the affected loop, while the logical switch can still work in a single-ring state with a packet throughput rate of 16Gbps.

The switches are stacked in daisy chain mode. Please refer to the figure below for the connection method.

Stacking increases the stability of switch ports and bandwidth.

4. Hot Standby (HSRP)

The core switch is the heart of the entire network. If a fatal failure occurs in the core switch, the local network will be paralyzed, and the loss caused is difficult to estimate. Therefore, when we choose a core switch, we often see that some core switches have stacking or hot backup functions.

Using hot backup for core switches is an inevitable choice to improve network reliability. When a core switch fails to work, all its functions will be completely taken over by another backup router in the system until the problematic router returns to normal. This is the hot backup routing protocol.

The condition for implementing HSRP is that there are multiple core switches in the system, which form a "hot standby group" that forms a virtual router.

At any time, only one router in a group is active and forwards data packets. If the active router fails, a backup router will be selected to replace the active router. However, from the perspective of the hosts in this network, the virtual router has not changed.

Therefore, the host remains connected and is not affected by the failure, which better solves the problem of core switch switching.

In order to reduce network data traffic, after the active core switch and the backup core switch are set, only the active core switch and the backup core switch send HSRP packets regularly.

If the active core switch fails, the backup core switch will take over as the active core switch. If the backup core switch fails or becomes the active core switch, another core switch will be elected as the backup core switch.

When a line from an access layer switch to the main core switch fails, the data flow is switched to the backup machine.

The data link of access layer switch 1 is switched to core switch B, but access layer switch 1 loses 6 data packets during the switch, as shown in the figure above.

When the main link between the server and core switch A fails (such as lines, network cards, etc.), 6 data packets will be lost when the server's main network card switches to the backup network card. However, when the main link is restored, the server will automatically switch from the backup network card to the main network card, and no data packets will be lost during this switch.

The specific data flow of terminal access to the server is shown in the figure below.