Convenient multi-user access using the hierarchical connection and extension mode

Cascading Scaling

The cascade expansion mode is the most common and direct extension mode. Some earlier networks are built with HUB) as the cascade devices. Because the hub was already very expensive at that time, most enterprises could not choose a switch as a level-1 device. This is because most working group users require access, which is generally connected from a port on the hub to the dashboard. In this way, the access capability has been greatly improved. However, due to some interference and human factors, the overall performance is very low and only needs to be met by multiple ports, there is no time to consider the forwarding and switching function. Considering the forwarding performance and port attributes of different vswitches, the current cascading expansion mode can easily achieve multi-user access through a certain Topology Design.

The hierarchical connection mode is the most ideal way to build a large LAN. It can combine various Topology Design Technologies and redundancy technologies to achieve a hierarchical network structure. For example, redundant Topology Design through dual-homing, the Link Aggregation technology is used to achieve redundancy and Up Link bandwidth expansion. These technologies are now very mature and widely used in various LAN and man networks.

The cascade mode uses common Ethernet ports for hierarchical interconnection, such as 100MFE port, GE port, and emerging 10GE port.

The cascade mode is a mainstream technology in Ethernet extended port applications. It uses a unified network management platform to manage devices across the network, such as topology management and fault management. The hierarchical connection mode is also facing challenges. When there are many layers and a large convergence ratio exists between layers, Edge Nodes experience a lot of switching and caching, there will be a certain latency. The solution is to converge upstream ports to reduce the convergence ratio, improve the performance of the Upper-end devices, or reduce the level of connection. In the level-4 connection mode, we recommend that you do not have more than four layers to Ensure network efficiency. If the edge node of the network has a port that is extended by a radio Ethernet device such as a HUB, because it is in a pass-through mode, there is no switch and it is not included in the hierarchy. However, note that, in the CSMA/CD mechanism that the HUB works on, the impact on network performance caused by conflicting delivery may be much greater than that caused by SWITCH-level connections.

The hierarchical connection mode is an inevitable choice for building a structured network. The hierarchical connection uses generic cable optical fiber. Each component can be placed in any position, which is very conducive to integrated wiring.

Stack technology expansion

Stack technology is another type of technology that currently uses a large number of port extensions on Ethernet switches. It is a non-standard technology. Different vendors do not support hybrid stack. The stack mode is determined by each vendor and does not support the topology structure. Currently, there are two popular stacking modes: Chrysanthemum chain mode and star mode. The biggest advantage of stack technology is that it provides simplified local management and manages a group of switches as an object.

Chrysanthemum chain Stack

Chrysanthemum chain stack is a stack Technology Based on the cascade structure. It has no special requirements on the switch hardware. It supports port concatenation and software at a relatively high speed, finally, we can build a multi-switch cascade structure. Through the loop, we can achieve redundancy to a certain extent. However, in terms of exchange efficiency, the same-level connection mode is at the same level. The Chrysanthemum chain stack usually uses one high-speed port and two high-speed ports. The structure of the two is shown in. In a high-speed port GE) mode, the uplink and downlink of the same port are respectively sent and received, forming a ring structure. data exchange between any two member switches takes one week, the switching ports of all vswitches have low efficiency, especially when many layers are stacked, stacking ports will become a serious system bottleneck. Two high-speed ports are used to implement chrysanthemum chain stacking. Because more high-speed ports are occupied, you can choose to implement ring redundancy. Compared with the cascade mode, the chrysanthemum chain stack mode does not have topology management and generally cannot be distributed. It is applicable to single-node organizations that require high-density ports and can be used on the edge of the network.

Due to the need to eliminate the broadcast storm caused by the loop, the chrysanthemum chain structure is normally at any time, A route entry from a vswitch to a master vswitch can only use one high-speed port, that is, one high-speed port cannot share the uplink data pressure of the vswitch. All upstream switches must exchange data.

Chrysanthemum chain stack is a type of simplified stacking technology. It is mainly an extended port technology that provides centralized management, for the forwarding efficiency between multiple switches, the efficiency of the Single Port mode will be much lower than that of the cascade mode), the hardware must provide more high-speed ports, and the software must implement up link redundancy. Generally, the number of layers of chrysanthemum chain stacks should not exceed four layers. all Group members must be placed close enough to the same rack ).

The star stack technology is an advanced stack technology. For switches, an independent or integrated high-speed switching center stack center is required ), all stacked hosts are uplinked to a unified stack center through dedicated or general high-speed ports. The stack center is generally a dedicated ASIC-based hardware switching unit, based on its switching capacity, the bandwidth is generally between 10-32 GB, and its ASIC switching capacity limits the number of layers in the stack.

Star Stack

The star stack technology reduces the number of switches of all Stack Group members to a level of Matrix in the stack center. forwarding between any two end nodes only requires three exchanges, the forwarding efficiency is the same as the communication structure of edge nodes in the first-level connection mode. Therefore, compared with the chrysanthemum chain structure, it can significantly increase the data forwarding rate between stacked members, provides a unified management mode. A group of switches can appear as a single node in network management.

The star stack mode is applicable to a single-node LAN that requires high-efficiency and high-density ports. The star stack mode overcomes the high latency impact of multi-layer forwarding in the chrysanthemum chain stack mode, but needs to provide a high-bandwidth Matrix, the cost is high, and the Matrix interface is generally not universal. The stacked ports of both the stack center and the member switch cannot be used to connect to other network devices. Using a highly reliable and high-performance Matrix chip is the key to star stack. Generally, the bandwidth of stacked cables is two-way between 2G and 2g), slightly higher than that of general GE. The higher part is usually used only for member management, so the effective data bandwidth is basically similar to GE. However, due to the dedicated bus technology, the cable length generally cannot exceed 2 m. Therefore, in star stack mode, all switches must be limited to one rack.

It can be seen that the traditional stack technology is a centralized management port expansion technology, which cannot provide topology management, has no international standards, and has poor compatibility. However, in a single-node LAN that requires a large number of ports, star stack provides excellent forwarding performance and convenient management features. Level-based connections are the basis for building networks. They can use various topologies and redundancy technologies flexibly. When there are too many layers, they need to be carefully designed. For networks with few levels of connection, the connection mode can provide optimal performance. For example, in a dot that needs to be expanded to double the port number, the edge of a star stack needs to be exchanged three times, and the cascade mode and the chrysanthemum chain stack need to be exchanged twice, the star stack mode requires a larger investment, and the chrysanthemum chain stack mode requires more high-speed ports. The common cascade mode is the most economical and efficient way to build. In addition, you can also use existing exchange devices without repeated investment. However, these two devices need to be managed separately.

Traditional stacking technology applications are often limited by geographical locations and often need to be placed in the same rack. In high-density port applications, wiring is difficult. Therefore, major manufacturers are actively seeking to support the distributed stack technology. At present, both Huawei's Quidway S Series Ethernet switches and Cisco Series Ethernet switches provide cluster management modes. Quidway S Series Ethernet switches use Huawei's unified VRP operating system and unified iManager network management system. The network management system supports the Chinese interface and adopts standard protocols and open technology, fully compatible with mainstream network management platforms. The Quidway S Series Ethernet switches are guided by the design philosophy of "four full-line speeds" under the guidance of Huawei's two-layer switching full-line speeds, three-layer switching full-line speeds, service switching full-line speeds, and QoS full-line speeds, make full use of the latencies of product development to meet the needs of broadband metro networks and enterprise networks in terms of product system design, scalability, and rich service features, it can provide customers with more efficient, secure, and scalable customer solutions.

Taking Huawei's HGMP product as an example, with the support of the cluster management mode, you can achieve centralized configuration and management on the network built using the Quidway series switch through the general level-level connection mode, a lan can be added to a group. For a network management system, a group can act as the same device and manage with an IP address, which is equivalent to or even better than the management effect of the previous Stack Group. However, as a universal centralized performance, group members and switches can implement Topology Design and distributed placement of members in the group, and stack ports can be any universal port supported by devices or aggregation of ports used, this allows you to flexibly control the switching network stack bandwidth to meet higher flexibility requirements.

For different environments, the effects of different port expansion modes are inconsistent. In the current situation, the common cascade mode is the main application method for hierarchical networks. The star stack mode is a simple management mode that provides single-node port expansion, the distributed stack implemented through cluster management will be the main method of the Next Generation stack.