Switch Cascade, stacking, cluster technology introduction (reprint)

The main port on the switch usually has a Uplink port and other common ports. Typically, the connection between the switches can be connected by Uplink port, may be connected to the ordinary port. Uplink employed to connect, also includes two cases. One case Uplink Uplink port connected to a port of another switch; Another is connected to the common port Uplink port of another switch. However, different connection, the requirements for the cable are different. Typically, the port used to connect Uplink (Uplink end use of the interface, the other end in common interface) between the switch and the switch is recommended. Because in this case, it is possible to use straight lines to connect. If two switches are used Uplink port, then the connection must be performed using a crossover cable. And this connection is to customers now using Uplink two switches directly connected to the interface, and use is indeed a direct line. This leads to two switches can not communicate.

 

Multi-switch in the LAN environment, the cascade switch, stack and cluster are the three important technologies.

Cascade technology to interconnect multiple switches;
stacking techniques may be composed of a plurality of unit switches, thereby improving the port density greater performance and higher;
clustering techniques can be connected to each other as a plurality of switches logical devices are managed, thereby greatly reducing the cost of network management, simplifying management operations.

1. Cascade

Cascade may be defined as two or more than two switches interconnected by a certain way.

MAN is an excellent example of switch cascading. Currently the telecommunications sector has been built around a number by Municipal broadband IP MAN. These broadband metropolitan area from the top down is generally divided into three layers: the core layer, convergence layer and access layer. The core layer typically uses Gigabit Ethernet technology, the convergence layer 1000M / 100M Ethernet technology, access layer 100M / 10M Ethernet technology, the so-called "Gigabit to the building, Fast to the floor, ten trillion to the desktop."
This broadband MAN is actually kind of structure of each of the many levels of cascading switches. The core switch (or router) aggregation switch connected a number of units, a plurality of second line aggregation switch cell center stage switch, the cell center stage switch even several premises switch even under a number of building floors switch station (or cell) switch (or hub).

Inter-switch is generally carried out by ordinary cascade port users, some switches provides a dedicated cascade port (Uplink Port). The two ports only difference is that ordinary MDI port meet standards, and the cascade port (or uplink port) MDIX compliance standards. Thereby resulting in a different embodiment of the connection of two ways: when the two switches are cascaded through common port, straight-through cable between port cable (Straight Throurh Cable); and only one of them through the cascade ports using crossover cable (crossover cable).
Here Insert Picture Description
Here Insert Picture Description

Pay attention to the following questions with a cascade switch:

  1. In principle, any manufacturer, any type of Ethernet switches can be cascaded, but does not rule out some special cases two switches can not be cascaded.
  2. Inter-switch cascading layers there is a certain limit. The most successful cascade of more of this principle is that the distance between any two sites can not exceed the maximum span media segment.
  3. When cascading multiple switches, should ensure that they all support the spanning tree (Spanning-Tree) protocol, is necessary to prevent loops in the network, but also to allow redundant links exist.
  4. Lower overall speed of the switch, the bandwidth is limited to an upper port of the switch. Subordinate equipment can only share a narrow exit. When cascaded, we should try to ensure that the relay link between the switches has sufficient bandwidth, may be employed for this technique and full-duplex link aggregation technology.

2. stack

Stacking means one or more switches in combination work together to provide as many ports in a limited space.
Here Insert Picture Description

The advantage of the stack

Expansion port stackable switches is the quickest and most convenient way.
● high port density
different brands of switches support different layers of the stack, in general, at least two layers may be stacked, and the stacking of up to 8 layers, thus, can provide a dense network of computers in a small space hundreds of ports.
● speed transmission
communication between the computer since the stack all the computers connected to the same high-speed backplane module, located in different layers of the switch ports is no longer required to forward, to reduce the forwarding delay between switches, to avoid port conflicts, among all the computer ports are wire-speed can be exchanged to improve the communication speed switch between different computers.
● easy to manage
a plurality of stack switches may be considered as a switch management table, which gives only an IP address to manage all the switches, the IP address, thereby greatly reducing the strength and difficulty of management , greatly saving management costs.

Rack switches can be said to be stacked higher stage of development to get the product. Rack Switches are generally above the department level have to switch, which has a plurality of slots, a large port density, support a variety of network types, scalability, good, strong processing capability, but expensive.

And 3. the difference between the stacked cascaded

Simply put: the stack is equal relationship, and the cascade is the superior-subordinate relationship.

Indeed the stack of each switch connected together motherboard bus, the delay between any two ports of different switches are equal, that is, the delay of a switch. The cascade will have a relatively large delay (on the lower level of the cascade relationship). Cascade level is limited. And performance of each layer is different, the worst performance of the final layer. Stacked backplane bandwidth is to share all the stack switches. For example, a switch backplane bandwidth of 2G, then 3, then the stack switches, each switch backplane bandwidth of the exchange, there 6G. And the stack is at the same level relations, the performance of each switch is the same.

Stacked and cascaded Contact

The stack can be seen as a special form of cascade. They differ in that: can be physically distant between the cascaded switches (permissible range within the media), while the distance between the multiple switches within a very close stacking unit, generally not more than a few meters; cascade commonly used common port, typically using a dedicated stacked and stacked module stacked cables.

In general, different manufacturers, different types of switches can be cascaded to each other, stacking is different, it must be the same type of stack switches between (or at least should be of the same manufacturer switches); cascade only between switches simple connection, the entire stack sucked stacking unit may be used as a switch, which not only means that the port density increases, but also means that the system bandwidth is broadened.

At present, the main switch on the market can be divided into stackable and non-stacked two categories. The so-called stackable switch, there are virtual stack and the stack of real points. The so-called virtual stack, is in fact a cascading between switches.

Not switch, but by stacking Fast Ethernet ports or Giga Ethernet port via a dedicated cable stacking module and stacked, which is in fact a disguised form of a cascade. Even so, a virtual stack has multiple switches can be managed as a logical device in the network, so that network management becomes easy for them.

The true sense of the stack should be satisfied: the use of special stacking modules and stacking stack bus does not occupy network interface; stack multiple switches, having a system bandwidth enough to ensure wire-speed switching each port can still reach the stack; multiple after the stack switches, VLAN and other functions unaffected.

Currently on the market there are a considerable number of stackable switches are virtual rather than real stacked type stacked type. Obviously, the stack than the true sense of the virtual stack on a much higher performance, but uses a virtual stack of at least two benefits: a virtual stack often using standard Fast Ethernet or Giga Ethernet as a stack bus, easy to implement, low cost; stacking ports can be used as ordinary port to use, help protect user investment. Standard Fast Ethernet ports or Giga Ethernet virtual stack, the stack can greatly extend the range of such a stack is no longer confined within the enclosure.

Clusters

The so-called clustering, is a plurality of interconnected (cascaded or stacked) of the switch as a logical device management.

Clusters, in general, only one played the role of managing switch, called a command switch, it can manage a number of other switches. In the network, these switches only need to use an IP address (only command switches need), saving valuable IP address. Under the unified management command switch, the cluster multiple switches to work, greatly reducing management intensity. For example, an administrator need only version can be upgraded all the switches in the cluster by the command switch.

The benefits of clustering technology to network management brings is beyond doubt. But to use the technology, it should be noted that different manufacturers have different cluster implementations, manufacturers generally are based on a proprietary protocol cluster. This determines the cluster technology has its limitations. The switches can be cascaded different manufacturers, but not the cluster. Even the same manufacturer's switch, only the specified model to achieve the cluster.

Differences and connections cascade, stack and cluster

Cascaded switches, stack, clusters of these three different but related technologies.

  • Cascading and stacking is a prerequisite for the cluster, the cluster is cascading and stacking purposes;
  • Cascading and stacking is a hardware-based implementation; clustering software-based implementation;
  • Cascading and stacking sometimes very similar (especially the cascade and virtual stack), and sometimes very different (and true cascade stacking).

Reference: https://blog.51cto.com/skypegnu1/1565537

Guess you like

Origin www.cnblogs.com/eos666/p/11955651.html