Basic vswitch knowledge: Distributed Link aggregation technology

Vswitch basics: aggregation technology in IRF Architecture

IRF technology combines multiple switching devices into a high-performance whole to achieve the highest possible network performance and network availability with minimal overhead. Devices supporting IRF technology have three important features: distributed device management (DDM) and Distributed Link aggregation.DLAAnd distributed elastic routing (DRR. These three technologies are indispensable for achieving IRF technical goals. DLA is used to improve the availability and capacity of the transmission link.

After multiple IRF switches are stacked, the number of ports increases, requiring DLA to support more aggregation groups, with each group having moreLink AggregationMember. More aggregation groups mean that the switching device can provide more high-speed links, while more aggregation members can not only increase the link capacity, but also reduce the risk of failure of the entire data line. The two parameters are different on different IRF devices, but the IRF system supports at least eight aggregation links. Each group can provide a transmission link with a total capacity of 80 M, M, or M. Some vswitches with higher configurations also allow aggregation of two 10g ports to provide a link with higher bandwidth.

In addition to providing greater bandwidth, DLA also achieves the aggregation goal in the IEEE 802.3ad standard:

1. The increase in bandwidth is controllable and linear, and can be determined by the user's configuration, not increasing in multiples of 10.

2. DLA automatically distributes traffic to aggregate members based on the data content to achieve load balancing.

3. Aggregate members dynamically back up each other. failure or replacement of a single link will not cause Link failure.

4. The upper-layer applications that use the service are transparent to the selection and replacement of the working links in the aggregation.

5. When the link connection or configuration parameters of the switching device change, DLA quickly calculates and re-sets the aggregation link to minimize the interruption time of the data stream.

6. If you have not manually set an aggregation link, the system can automatically set an aggregation link to bind the physically matched links.

7. The results of Distributed Link aggregation are predictable and definite. They are only related to the link parameters and physical connections, and are irrelevant to the parameter configuration or change sequence.

8. No matter whether the aggregation link is stable or re-converged, the data sent and received will not be duplicated or disordered.

9. It can communicate with switches that do not support aggregation technology, and can also communicate with other devices that support aggregation technology.

10. You can configure aggregation parameters or view the aggregation status through CONSOLE, SNMP, TELNET, and WEB.

Vswitch basics: DLA features

As a new technology, IRF technology presents many new features. Its distributed architecture makes its functions distinctive. DLA demonstrates the uniqueness of IRF in link aggregation:

1. Support for non-continuous port Aggregation

Unlike the previous aggregation implementation method, the IRF system does not require the members of the same aggregation group to be the ports consecutively numbered on the device. Any data port can be aggregated as long as certain aggregation conditions are met. You can flexibly build an aggregation link based on the available ports on the current switching system.

2. Supports cross-device and cross-chip Aggregation

Currently, some stack technologies do not support cross-device aggregation. That is, only ports on the same physical device in the stack can be added to the same aggregation group. You cannot specify aggregation members at will. To some extent, these restrictions offset the benefits of port number expansion. For example, if a user wants to increase the capacity of a transmission line to MB through aggregation, this requirement cannot be met if there are less than 8 ports on each individual device. Although the entire system still has enough M ports available, they are scattered across physical devices and cannot form a logical link meeting bandwidth requirements.

Vswitch basics: IRF differences

In IRF's view, multiple devices on the stack are called units. Link aggregation functions and operations should also be a whole. The DLA module shields users from the specific physical location of the port. For details, see Figure 3. As long as the aggregation conditions are the same, you can aggregate the ports of different units. The ports p1, p2, p3, and p4 constitute a logical link. Now, unit1 ~ 4. collaborative computing and selection of work links in an aggregation group. P1 ~ P4 dynamically backs up each other to enable cross-device data sending and receiving and load balancing, maximizing the advantages of multiple devices.

Cross-device aggregation links

In addition, some exchange devices do not support cross-chip aggregation, that is, ports on different exchange chips of the same device cannot be aggregated. This restriction does not exist for IRF devices. DLA allows ports to form an aggregation group across chips. For some IRF devices that use daughter cards, the ports on the daughter cards can also be aggregated with any matching port in the current unit or other units.

Vswitch basics: Distributed Link aggregation Control

Although the IRF system is presented as a whole, it does not limit that users can only operate on a specific unit. Taking aggregation as an example, you can configure and manage all aggregation links on any unit of the system to view the status of all aggregation groups and aggregation ports. You can connect to any unit of the system through CONSOLE, SNMP, TELNET, or WEB. Then, you can create or delete an aggregation group to display the aggregation information, you can also enter the specific port mode to modify or display its aggregation parameters. In this process, DLA automatically sends USER commands to the unit of the port for Synchronous execution. The unit that receives the command gets the execution result and provides it to the user.

The distributed aggregation technology further eliminates device spof and improves link availability. Since the aggregation members can come from different devices, even if some units in the system fail, other normally working units will continue to control and maintain the status of the remaining ports, and the aggregation link will not be completely interrupted. This is of great significance to core exchange systems and networks with high-quality services. The following figure 4 shows an aggregation link between the IRF system X1 and X2. The physical connection of this link is Link1 ~ Link4 is responsible for communication between LAN1 and lan2. If the X11 switch in X1 fails, the Link1 and Link2 are unavailable, and Link3 and Link4 are not affected, data can still be sent and received together. After that, if X22 in X2 fails, the connection between X1 and X2 can be maintained at least through Link3.

Aggregation link between two IRF Systems

An IRF device can be considered as a "modular" scalable) switch. You can use a single IRF switch to build a network, or add one to enhance the performance of network devices as needed. At the same time, this high-performance stack switch can also be split. After the split, each unit becomes an independent switching device. The above process is called merge) and split ). If the first two systems have created the same aggregation link with the same parameters, IRF requires that the aggregated members be added to the same group, that is, the aggregation group is also merged. After merging, each unit works collaboratively. It matches configuration parameters globally, assigns aggregation group numbers, adds ports to the corresponding group, and recalculates and sets the port status. Similarly, if the members of the same aggregation group are distributed on different units before splitting, they are still in the same type of aggregation group created by each group. DLA ensures that each unit retains the current aggregation configuration, deletes the ports that have been left from the group, and then calculates the status of the remaining ports.

This feature maximizes the protection of user aggregation configurations. In addition, when the stacked link failure causes System Splitting, this feature allows the IRF system to maintain the existing aggregation link as much as possible, reducing the data transmission loss caused by the failure.

Vswitch basics: Multiple aggregation types

DLA implements three types of aggregation methods: manual aggregation, static aggregation, and dynamic aggregation.

Manual and static aggregation groups are created or deleted using user commands. The group members are also specified by users. After the group is created, the system cannot automatically delete the group or change the Group members. However, you must calculate and select the group members. Whether an aggregate member becomes a work link depends on its configuration parameters. Not all members can participate in data forwarding.

Manual and static aggregation are mainly different in the aggregation control mode. LACP is not enabled on the manual aggregation link, and configuration information is not exchanged with the peer system. Therefore, the aggregation control only determines the work link based on the system configuration. This aggregation control method is more common on earlier exchange devices. Static aggregation groups are different. Although the aggregation members are specified by users, DLA automatically starts LACP on the static link. If LACP is enabled for the peer system, both devices can exchange aggregated information for the aggregation control module.

Dynamic aggregation control fully complies with the LACP protocol and achieves the goal of automatic configuration of aggregation links in the 802.3ad standard. You only need to select a dynamic mode for the port, and the system will automatically aggregate the port matching parameters to set its working status. In the dynamic aggregation mode, the system sends LACP packets to each other and exchanges status information to maintain aggregation. If the parameter or status changes, the link is automatically removed from the original aggregation group and added to another suitable group.

The above three aggregation methods provide good aggregation compatibility for the IRF system. The system can not only interconnect devices that do not support link aggregation, but also work with devices of different aggregation implementations. You can flexibly select the aggregation type based on the actual network environment to achieve high-performance and high-reliability links.