Introduction to various technologies in core Switches

As an indispensable device in networking, core switches are gradually improving their functions to meet more user needs. The simplest LAN is usually composed of a hub (or switch) and several computers. As the number of computers increases and the network size expands, in more and more LAN environments, core switches replace hubs, and multiple switches replace single switches.

In multi-switch LAN environments, switch cascade, stacking, and cluster are three important technologies. Cascade technology can interconnect multiple switches. stacking technology can combine multiple switches into one unit to increase the port density and performance; the cluster technology manages multiple interconnected core switches as one logical device, greatly reducing network management costs and simplifying management operations. Considering the development status of local area network, the local area network (LAN) mentioned in this Article, if not specified, refers to 10 BaseT, 100 BaseT (F), 1000 BaseT (F) switched Ethernet.

I. Cascade

Cascade switches can be defined as two or more core switches connecting each other in a certain way. As needed, multiple core switches can be cascade in multiple ways. In a large LAN, such as a campus network (Campus Network), multiple switches form a hierarchical structure of bus, tree, or star based on their performance and purpose.

Man is an excellent example of core switch cascade. At present, China Telecom has built many municipal-level broadband ip man regions. These broadband MAN networks are generally divided into three layers: core layer, convergence layer, and access layer. The core layer generally uses Gigabit Ethernet technology, the convergence layer uses 1000 M/100 M Ethernet technology, and the access layer uses M/10 M Ethernet technology, the so-called "Gigabit to the building, MB to the floor, 10 MB to desktop ". This structure of the broadband MAN is actually formed by the cascade of many switches at various levels. A core switch (or router) is connected to several aggregation switches. The aggregation switch is associated with several residential center switches, and the residential center switch is connected to several building switches, A building switch is connected to several floor (or unit) switches (or hubs ).

Generally, switches are cascade through common user ports. Some switches provide dedicated Uplink ports ). The difference between the two ports is that normal ports comply with the MDI standard, and cascade ports (or uplink ports) comply with the MDIX standard. As a result, two different connection modes are achieved: When both switches are connected through a common port cascade, the inter-port Cable uses a Straight-through Cable (Straight Throurh Cable ); crossover Cable is used when only one of them uses cascade ports ). To facilitate cascade, some switches provide a dual-purpose port, which can be set to MDI or MDIX through switches or management software. Furthermore, all or part of the ports on some switches have the MDI/MDIX self-calibration function, which can automatically differentiate the network cable type for more convenient cascade.

Pay attention to the following issues when using vswitches for Cascade. In principle, any manufacturer or any type of Ethernet switch can be cascade, but it does not rule out that two switches cannot be cascade in some special circumstances. The number of cascade layers between vswitches is limited. The most fundamental principle for successful cascading is that the distance between any two sites cannot exceed the maximum span of the media segment. When multiple switches are cascaded, make sure that they all support the Spanning Tree Protocol, which not only prevents loops in the network, but also allows redundant links.

During cascade, we should try our best to ensure that the Relay Links between switches have sufficient bandwidth. Therefore, we can adopt full duplex and link aggregation technologies. After the full-duplex technology is adopted for the core switch ports, the throughput of the corresponding ports is doubled, and the relay distance between the core switches is greatly increased, making it possible to cascade multiple switches with remote distribution and distance. Link aggregation is also called Port aggregation, port bundling, and link expansion combination. It is defined by the IEEE 3AD standard. That is, two devices connect to each other through more than two ports of the same type, and transmit data at the same time to provide higher bandwidth, better redundancy, and achieve load balancing. Link aggregation technology not only provides high-speed connections between switches, but also provides high-speed channels for connections between switches and servers. Note that not all types of switches support these two technologies.

Ii. Stack

Stack (stack in some books) means to combine more than one core switch to work together to provide as many ports as possible in a limited space. Multiple switches are stacked to form a stacked unit. The stackable switch performance indicator has a "Maximum stackable number" parameter, which refers to the maximum number of switches that can be stacked in a stack unit, represents the maximum port density provided by a stack unit.

The concepts of stack and cascade are both different and related. Stack can be seen as a special situation of cascade. The difference between them is that cascade switches can be far apart (within the scope of the media license), while the distance between multiple switches in a stacked unit is very close, generally, only a few meters are used. Generally, common ports are used for Cascade, while dedicated stacking modules and stacked cables are used for stacking. Generally, Switches of different manufacturers and models can be cascade with each other, but stacks are different. Switches of the same type (at least Switches of the same manufacturer) must be stacked; cascade is only a simple connection between switches. Stack uses the entire stack unit as a switch. This not only increases the port density, but also increases the bandwidth of the system.

Currently, mainstream switches on the market can be divided into two categories: Stack type and non-stack type. Vswitches that can be stacked are divided into virtual stacks and real stacks. The so-called virtual stack is actually the cascade between switches. The switch is not stacked through a dedicated stack module and stack cable, but through the Fast Ethernet port or Giga Ethernet port. In fact, this is a disguised cascade. Even so, multiple vswitches in the virtual stack can be managed as a logical device in the network, making network management easy. In the true sense, the stack should satisfy the following requirements: a dedicated stack module and a stack bus are used for stacking without occupying Network Ports. After multiple switches are stacked, the system bandwidth is sufficient, this ensures that each port can achieve line rate switching after stacking. After multiple switches are stacked, functions such as VLAN are not affected.

At present, a considerable number of Stackable switches on the market belong to the virtual stack type rather than the real stack type. Obviously, the real stack performance is much higher than that of the virtual stack, but the use of virtual Stack has at least two advantages: standard Fast Ethernet or Giga Ethernet is often used as the stack bus for virtual stack, which is easy to implement and cost-effective. Stack ports can be used as common ports to protect users' investment. The use of standard Fast Ethernet or Giga Ethernet ports for virtual stacking can greatly extend the stack range, so that the stack is no longer limited to one cabinet.

Stack can greatly improve the port density and performance of the core switch. The stacking unit is sufficient to match the port density and performance of a large rack-mounted switch, while the investment is much lower than the rack-mounted switch, and the implementation is much more flexible. This is the advantage of stack. Rack-mounted switches are the product of the development of stacks to a higher stage. A rack-mounted switch is generally a department-level switch. It has multiple slots and has a large port density. It supports multiple network types, provides good scalability and processing capabilities, but is expensive.

Iii. Clusters

A cluster manages multiple switches that are connected to each other (cascade or stack) as a logical device. In a cluster, there is generally only one core switch for management, called a command switch. It can manage several other switches. In the network, these switches only need to occupy one IP address (only required by the command switch), saving valuable IP addresses. Under the unified management of command switches, multiple switches in the cluster work collaboratively, greatly reducing the management intensity. For example, the Administrator only needs to use the command switch to upgrade all the vswitches in the cluster.

The benefits of cluster technology to network management are beyond doubt. However, to use this technology, we should note that different manufacturers have different implementation solutions for clusters. Generally, the manufacturers use proprietary protocols to implement clusters. This determines that the cluster technology has its own limitations. Core Switches of different manufacturers can be cascaded, but cannot be clustered. Only the specified model of the core switch of the same manufacturer can implement the cluster. For example, the CISCO 3500XL series can only implement clusters with the 1900, 2800, and 2900XL series.

The cascade, stack, and cluster technologies of vswitches are different and related. Cascade and stack are the prerequisites for implementing clusters. clusters are the purpose of cascade and stack. cascade and stack are implemented based on hardware. clusters are implemented based on software; cascade and stack are sometimes very similar (especially cascade and virtual stack), sometimes very different (cascade and real stack ). With the development of LAN and man, the above three technologies will be applied more and more widely.