Classification of switches based on management capabilities. Workgroup Switches

General classification of switches

Computer A network is a group of computers connected to each other by a communication channel. The channel ensures data exchange within the network, that is, data exchange between computers of a given group. The network can consist of two or three computers, or it can unite several thousand PCs. Physically, data exchange between computers can be carried out via a special cable, fiber optic cable or through twisted pair.

Network hardware and hardware-software help connect computers into a network and ensure their interaction. These funds can be divided into the following groups according to their main functional purpose:

Passive network equipment connecting connectors, cables, patch cords, patch panels, telecommunication sockets, etc.;

Active network equipment converters/adapters, modems, repeaters, bridges, switches, routers, etc.

Currently, the development of computer networks occurs in the following areas:

Speed increase;

Implementation of switching-based segmentation;

Connecting networks using routing.

Layer 2 Switching

Considering the properties of the second layer of the ISO/OSI reference model and its classical definition, we can see that this level belongs to the main share of commuting properties.

The data link layer ensures reliable transit of data across a physical channel. In particular, it addresses issues of physical addressing (as opposed to network or logical addressing), network topology, line discipline (how the end system should use the network link), fault notification, ordering of data blocks, and information flow control.

In fact, the functionality defined by the OSI data link layer serves as the platform for some of today's most powerful technologies. The importance of Layer 2 functionality is underscored by the fact that hardware manufacturers continue to invest heavily in developing devices with such functionality, i.e. switches.

Layer 3 switching

Layer 3 switching? This is hardware routing. Traditional routers implement their functions using software-controlled processors, which we will call software routing. Traditional routers typically forward packets at a rate of about 500,000 packets per second. Layer 3 switches today operate at speeds of up to 50 million packets per second. It is also possible to further increase it, since each interface module, as in the second level switch, is equipped with its own ASIC-based packet forwarding processor. So increasing the number of modules leads to increasing routing performance. Usage high speed technology Large custom integrated circuits (ASIC) is main characteristic which differentiates Layer 3 switches from traditional routers.

A switch is a device that operates at the second/third level of the ISO/OSI reference model and is designed to combine network segments operating on the same link/network layer protocol. The switch routes traffic through only the one port needed to reach its destination.

The figure (see Figure 1) shows the classification of switches according to management capabilities and in accordance with reference model ISO/OSI.

Figure 1 Switch classification

Let's take a closer look at the purpose and capabilities of each type of switch.

Unmanaged switch? This is a device designed to connect several nodes computer network within one or more network segments. It transmits data only directly to the recipient, with the exception of broadcast traffic to all network nodes. An unmanaged switch cannot perform any other functions.

Managed switches are more complex devices that allow you to perform a set of functions of the second and third levels of the ISO/OSI model. They can be managed via the Web interface, command line via the console port or remotely via SSH, as well as using the SNMP protocol.

Configurable switches provide users with the ability to configure specific settings using simple management utilities, a Web interface, a simplified command line interface, and SNMP.

Layer 2 switches analyze incoming frames, decide on their further transmission, and forward them to destinations based on the OSI link layer MAC addresses. The main advantage of Layer 2 switches is transparency to upper-layer protocols. Since the switch operates at layer 2, it does not need to analyze information from the upper layers of the OSI model.

Layer 3 switches perform switching and filtering based on the addresses of the link (layer 2) and network (layer 3) layers of the OSI model. Such switches dynamically decide whether to switch (layer 2) or route (layer 3) incoming traffic. Layer 3 switches perform switching within working group and routing between different subnets or virtual local area networks (VLANs).

Send your good work in the knowledge base is simple. Use the form below

Students, graduate students, young scientists who use the knowledge base in their studies and work will be very grateful to you.

Posted on http://www.allbest.ru/

1. Classify Switchs on technological implementation

LAN switches come in a wide variety of features and prices.

One of the reasons for such large differences is that they are intended to solve various classes tasks. High-end switches should provide high performance and port density and support a high range of management functions. And lower-class switches usually have a small number of ports and are not able to support management functions.

One of the main differences is the architecture used in the switch:

1. Based on the switching matrix (cross-bar);

2. With shared multi-input memory (shared memory);

3. Based on common high-speed bus.

Often these three communication methods are combined in one switch.

2. Classify switches by design

1. Standalone switches with a fixed number of ports;

2. Modular chassis-based switches;

3. Switches with a fixed number of ports, assembled into a stack.

3. Classifyswitches by operating level

Depending on the level at which the switch operates, switching is divided into switching of the 2nd, 3rd and 4th levels.

1. Layer 2 switching - hardware. There are 2 main reasons for using Layer 2 switches - network segmentation and workgroup aggregation;

2. Layer 3 switching - decisions are made based on network layer information, and not based on MAC addresses. The main purpose of Layer 3 switching is to achieve Layer 2 switching speed and routing scalability;

3. Layer 4 switching - the decision to transmit a packet is based not only on MAC or IP addresses, but also on Layer 4 parameters, such as the TCP/UDP port number.

4. Give an excellentSwitch connection from hub

1. Network Scalability - In a network built on hubs, bandwidth is shared, thereby limiting the bandwidth of each node and making it very difficult to grow the network without losing performance.

2. Latency - the amount of time it takes for a packet to reach its destination. Since each node in the network built on hubs must wait for the possibility of data transmission to avoid collisions, the delay can increase significantly as the number of nodes in the network increases.

Simply replacing hubs with switches can dramatically improve efficiency local networks, no replacement required

cabling or network adapters. Switches divide the network into separate logical segments, while creating separate, small-sized collision domains on each port. Dividing a large network into several autonomous segments using switches has several advantages:

1. Since only part of the traffic is redirected, switches reduce the traffic received by devices in all network segments;

2. All nodes connected to the hub share the entire bandwidth. Switches provide each node (if it is connected directly to the switch port) with separate bandwidth, thereby reducing the likelihood of collisions in network segments.

For example, if 10 devices are connected to a 10 Mbps hub, then each node will receive less than 1 Mbps throughput (10/N Mbps, where N is the number of workstations), even if not all devices are transmitting data. If you install a switch instead of a hub, then each node will be able to operate at a speed of 10 Mbit/s.

5. Give the main characteristics of switches that affect performance

The main indicators of the switch that characterize its performance are:

1. Frame filtering speed;

2. Speed of personnel promotion;

3. Bandwidth;

4. Frame transmission delay.

Additionally, there are several switch characteristics that have the greatest impact on these performance specifications. These include:

1. Size of the internal address table.

2. Size of frame buffer(s).

3. Switching type - “on the fly” or with intermediate storage.

4. Internal bus performance.

5. Performance of the processor or processors.

6. Describe the main types of connections to managed switches

Before you begin configuring the switch, you must establish a physical connection between the switch and the workstation. There are two types of cabling used to manage the switch. The first type is through the console port (if the device has one), the second is through the Ethernet port (via the Telnet protocol or via the Web interface).

For example, D-Link managed switches have a console port that connects to a computer serial port using the included RS-232 cable. Console connection is sometimes called ` Out- of- Band"connection. This means that the console is using a different network connection circuit (does not use the bandwidth of the Ethernet ports). It can be used to install and manage the switch even when there is no network connection.

7. Describe the main three typesVLAN

Switches allow you to implement three types of VLAN:

1. VLAN based on ports.

2. VLAN based on MAC addresses.

3. VLAN based on tags in the additional field of the frame (IEEE 802.1q standard).

8 . ThuTaggedone ofVLAN:

Tagging(Package marking) -the process of adding 802.1q VLAN membership information to the frame header. Ports on which packet tagging is enabled can add a VID number, priority information, etc. to the headers of all transmitted packets. If a packet arrives at a port already tagged, then this packet does not change and thus all VLAN information is preserved during forwarding. Packet tagging is primarily used to forward packets between devices that support the 802.1q VLAN standard.

9 . Thuo happens to a packet that hits the portUntaggedone ofVLAN

· Untagging -The process of extracting 802.1q VLAN information from the packet header. Ports on which this feature is enabled extract all VLAN-related information from the headers of both incoming and outgoing packets passing through the port. If the packet does not contain a virtual network tag, then the port does not modify such a packet. This function switch is used when transmitting packets from switches that support the 802.1q standard to devices that do not support this standard.

10 . OnThere are two main ways to create reliable communication channels using managed switches:

The most common is to create redundant connections between switches based on two technologies:

1. Redundancy mode, when one of the connections is functioning, and the rest are in “hot” standby to replace a failed connection.

2. Load balance mode; in this case, data is transmitted in parallel over all alternative connections. To implement the mode, port aggregation is used.

Consolidation (aggregation) of ports (Port Trunking) - it's unitedeconnection of several physical channels (Link Aggregation) into one logical mAgistral.

switch hub communication constructive

11 . KaWhat types of communication channel aggregation do you know:

Supports two types of link aggregation: static and dynamic.

With static link aggregation (set by default), all settings on the switches are performed manually.

Dynamic link aggregation is based on the IEEE 802.3ad specification, which uses the Link Aggregation Control Protocol (LACP) to check the link configuration and route packets to each physical link. In addition, the LACP protocol describes a mechanism for adding and removing channels from a single communication line. To do this, when configuring an aggregated communication channel on switches, the corresponding ports of one switch must be configured as “active” and the other switch as “passive”. "Active" LACP ports process and forward its control frames. This allows LACP-enabled devices to agree on aggregate link settings and be able to dynamically change the port group, i.e. add or exclude ports from it. “Passive” ports do not process LACP control frames.

The IEEE 802.3ad standard is applicable to all types of Ethernet channels, and with its help you can even build multi-Gigabit communication lines consisting of several Gigabit Ethernet channels.

12 . Onbasis on which the root switch is selected when constructing a tree according to the protocolSTP:

The STP algorithm requires that each switch be assigned an ID. The switch ID is an 8-byte field that consists of 2 parts: a 2-byte priority assigned by the administrator and a 6-byte MAC address of its control unit.

Each port is also assigned a unique identifier within the switch, usually its MAC address. Each switch port is assigned a route cost that corresponds to the cost of transmitting a frame over the local network through this port.

The process of calculating a spanning tree begins with choosing root switch (root switch), from which the tree will be built. AsTve of the root switch, the switch with the lowest value is selectedeID number.(Initially, by default, all switches have the same priority value of 32768. In this case, the root switch is determined by the lowest MAC address.) Sometimes, this choice may not be rational. To select as the root switch specific device(based on the network structure), the administrator can influence the election process by manually assigning the lowest ID to the corresponding switch.

The second stage of STP is selecting a root port for each of the remaining switches on the network.

The root switch port is the port that has the shortest distance across the network to the root switch.

The third step of how STP works is to determine the designated ports.

Each segment in a switched network has one designated port. This port functions as the only port on the switch, i.e. receives packets from the segment and forwards them towards the root switch through the root port of that switch.

The switch containing the designated port for this segmentAcalled a designated switch (designated bridge) of this segment. The designated port on a segment has the shortest distance to the root switch among all the ports connected to that segment.

A segment can only have one designated port. At the root switch, all ports are designated, and their distance to the root is set to zero. The root switch does not have a root port.

When constructing a spanning tree, the concept of distance plays an important role. This criterion selects a single port connecting each switch to the root switch, and a single port connecting each network segment to the root switch. All other ports are placed in a standby state, that is, one in which they do not transmit normal data frames. With this selection of active ports in the network, loops are eliminated and the remaining links form a spanning tree.

Posted on Allbest.ru

Types of switches

However, if previously managed switches differed from unmanaged ones, including a wider range of functions, now the difference can only be in the possibility or impossibility remote control device. For the rest - even at the most simple models Manufacturers add additional functionality, often increasing their cost.

Therefore on this moment The classification of switches by level is more informative.

Switch levels

In order to choose a switch that best suits our needs, we need to know its level. This setting is determined based on what OSI (data transfer) network model the device uses.

Devices first level, using physical data transmission have almost disappeared from the market. If anyone else remembers hubs, then this is just an example physical level when information is transmitted in a continuous stream.
Level 2. Almost all unmanaged switches fall into this category. The so-called channel network model. Devices divide incoming information into separate packets (frames), check them and send them to a specific recipient device. The basis for information distribution in second-level switches is MAC addresses. From these, the switch compiles an addressing table, remembering which MAC address corresponds to which port. They don't understand IP addresses.

Level 3. By choosing such a switch, you get a device that already works with IP addresses. It also supports many other possibilities for working with data: converting logical addresses to physical ones, network protocols IPv4, IPv6, IPX, etc., pptp, pppoe, vpn and others connections. On the third, network level of data transmission, almost all routers and the most “advanced” part of switches work.

Level 4. The OSI network model used here is called transport. Not even all routers are released with support for this model. Traffic distribution occurs at an intelligent level - the device can work with applications and, based on the headers of data packets, direct them to the desired address. In addition, transport layer protocols, such as TCP, guarantee the reliability of packet delivery, preservation a certain sequence their transmission and are able to optimize traffic.

Select a switch - read the characteristics

How to choose a switch based on parameters and functions? Let's look at what is meant by some of the commonly used symbols in specifications. Basic parameters include:

Number of ports. Their number varies from 5 to 48. When choosing a switch, it is better to provide a reserve for further network expansion.

Basic data rate. Most often we see the designation 10/100/1000 Mbit/s - the speeds that each port of the device supports. That is, the selected switch can operate at a speed of 10 Mbit/s, 100 Mbit/s or 1000 Mbit/s. There are quite a lot of models that are equipped with both gigabit and 10/100 Mb/s ports. Most modern switches operate according to the IEEE 802.3 Nway standard, automatically detecting port speeds.

Bandwidth and Internal Bandwidth. The first value, also called the switching matrix, is the maximum amount of traffic that can be passed through the switch per unit of time. It is calculated very simply: number of ports x port speed x 2 (duplex). For example, an 8-port gigabit switch has a throughput of 16 Gbps.
Internal throughput is usually indicated by the manufacturer and is only needed for comparison with the previous value. If the declared internal bandwidth is less than the maximum, the device will not cope well with heavy loads, slow down and freeze.

Auto MDI/MDI-X detection. This is auto-detection and support for both standards by which the twisted pair was crimped, without the need for manual control of connections.

Expansion slots. Possibility of connecting additional interfaces, for example, optical.

MAC address table size. To select a switch, it is important to calculate in advance the size of the table you need, preferably taking into account future network expansion. If there are not enough entries in the table, the switch will write new ones over the old ones, and this will slow down data transfer.

Form factor. The switches are available in two types of housing: desktop/wall-mounted and rack-mounted. In the latter case, the standard device size is 19 inches. Special ears for rack mounting can be removable.

We select a switch with the functions we need to work with traffic

Flow control ( Flow Control, IEEE 802.3x protocol). Provides coordination of data sending and receiving between the sending device and the switch under high loads, in order to avoid packet loss. The function is supported by almost every switch.

Jumbo Frame- increased packages. Used for speeds from 1 Gbit/sec and higher, it allows you to speed up data transfer by reducing the number of packets and the time for processing them. The function is found in almost every switch.

Full-duplex and Half-duplex modes. Almost all modern switches support auto-negotiation between half-duplex and full-duplex (transmitting data in one direction only, transferring data in both directions at the same time) to avoid problems in the network.

Traffic prioritization (IEEE 802.1p standard)- the device can identify more important packets (for example, VoIP) and send them first. When choosing a switch for a network where a significant portion of the traffic will be audio or video, you should pay attention to this function

Support VLAN(standard IEEE 802.1q). VLAN is a convenient tool for delimiting individual areas: internal network enterprises and networks common use for clients, various departments, etc.

To ensure security within the network, control or check the performance of network equipment, mirroring (traffic duplication) can be used. For example, all incoming information is sent to one port for checking or recording by certain software.

Port Forwarding. You may need this function to deploy a server with Internet access, or for online games.

Loop protection - STP and LBD functions. Particularly important when choosing unmanaged switches. It is almost impossible to detect the formed loop in them - a looped section of the network, the cause of many glitches and freezes. LoopBack Detection automatically blocks the port where a loop has occurred. The STP protocol (IEEE 802.1d) and its more advanced descendants - IEEE 802.1w, IEEE 802.1s - act a little differently, optimizing the network for a tree structure. Initially, the structure provides for spare, looped branches. They are disabled by default, and the switch only starts them when there is a loss on some of the main lines.

Link aggregation (IEEE 802.3ad). Increases channel throughput by combining multiple physical ports into one logical one. The maximum throughput according to the standard is 8 Gbit/sec.

Stacking. Each manufacturer has its own stacking design, but in general this feature refers to the virtual combination of multiple switches into one logical unit. The purpose of stacking is to get large quantity ports than is possible when using a physical switch.

Switch functions for monitoring and troubleshooting

Many switches detect a faulty cable connection, usually when the device is turned on, as well as the type of fault - broken wire, short circuit, etc. For example, D-Link provides special indicators on the case:

Protection against virus traffic (Safeguard Engine). The technique improves operating stability and protects CPU from overloads with “junk” traffic of virus programs.

Power Features

Energy saving.How to choose a switch that will save you energy? Pay attentione for the presence of energy saving functions. Some manufacturers, such as D-Link, produce switches with power consumption regulation. For example, a smart switch monitors the devices connected to it, and if any of them is not working at the moment, the corresponding port is put into “sleep mode”.

Power over Ethernet (PoE, IEEE 802.af standard). A switch using this technology can power devices connected to it over twisted pair cables.

Built-in lightning protection. Very required function, however, we must remember that such switches must be grounded, otherwise the protection will not work.

website

Back in the first issue of LAN magazine, in the “First Lessons” section, we published an article by S. Steinke “Ethernet switching” about the basics of this technology and we were not mistaken with our choice: over the next three years, Ethernet switching has become one of the “hottest” technologies. Later, we returned to this topic more than once (see, in particular, the article by D. Ganzhi “Switches in Local Networks” in the April 1997 issue of LAN). The first article appeared at a time when Fast Ethernet was still fighting for a place in the sun with 100VG-AnyLAN, and the outcome of the fight was far from clear, so it was devoted primarily to 10 Mbit/s switching. The second of these articles dealt mainly with general aspects of switching. Considering the above circumstances, as well as the importance of switching as such, we considered it possible and even necessary to return to this topic again, especially since the series of articles on Ethernet would not be complete without considering it.

WHAT IS A SWITCH?

A switch is essentially a multiport bridge, so like a bridge, it receives incoming packets, stores them temporarily, and then forwards them to another port according to the destination address of that packet. Switches can be used to connect different LANs, to segment a LAN (that is, reduce the number of nodes competing for media in the same collision domain), and to overcome restrictions on segment diameter. The latter application is especially important in the case of Fast Ethernet networks, where the segment diameter cannot exceed 205 m for twisted pair cable.

Switches use the concept of " virtual connection" to organize a temporary connection between the sender and the recipient. After transmitting the packet, the virtual connection is broken. The switch keeps a table where it remembers which stations (more precisely, which MAC addresses) are connected to which physical port. In Figure 1, the subscriber with address A sends the packet to the recipient with address D. Using the table, the switch determines that a station with address A is connected to port 1, and a station with address D is connected to port 4. Based on this data, it establishes a virtual connection to transmit a message between ports 1 and 4.

Picture 1.
Based on the recipient address, the switch determines which port to forward the incoming packet to.

In an Ethernet switch, data transfer between disjoint pairs of ports can occur simultaneously. For example, Host A might be sending a packet to Host D at the same time Host B is sending a packet to Host C. Both conversations are happening at the same time, so in the case of Ethernet, the total throughput of the switch in our example is 20 Mbps. It is determined by summing up the available bandwidth for each connection; for example, in the case of a 12-port Ethernet switch, it is theoretically equal to 60 Mbps. In comparison, an Ethernet repeater always has the same aggregate throughput of 10 Mbps, regardless of the number of ports. In addition, the actual throughput of a hub can be much lower when multiple devices are competing for access to the transmission medium. However, the actual total throughput of the switch may be lower than the theoretically calculated one due to shortcomings in the switch design, for example, due to inadequate internal bus throughput. In this case, the switch is said to have a blocking architecture.

SWITCH ARCHITECTURE

The architecture of a switch is determined by four main factors - port type, buffer sizes, packet forwarding mechanism, and internal bus (see Figure 2).

Figure 2.
With all the diversity in switch designs, the basic architecture of these devices is determined by four components: ports, buffers, an internal bus, and a packet forwarding mechanism.

Ports can have speeds of 10 and 100 Mbit/s and operate in half-duplex and full-duplex mode. Many high-end models may also contain FDDI, ATM, Gigabit Ethernet, etc. ports, but we will not touch on this topic here, especially since we have already briefly reviewed it earlier.

The presence of buffers of sufficient capacity has great importance for switching, in particular in the case of using sliding window protocols in the network, when the subscriber confirms the receipt of not each packet, but a series of them. Generally speaking, the larger the buffer capacity, the better, but the more expensive it is. Therefore, developers have to choose between performance and price. But they have another solution - thread control (see below).

The packet forwarding mechanism can be one of the following three: store-and-forward switching, cut-through switching, and hybrid cut-through switching. We have already looked at them several times, so let us just remind you what they are. In the first case, the packet is completely stored in a buffer before being transmitted further, so this method introduces the greatest delay, but also does not allow erroneous packets to leave the segment. In the second case, having read the recipient's address, the switch immediately transmits the frame further. As is easy to understand, it has exactly the opposite advantages and disadvantages - low latency and the lack of adequate frame checking.

In the third case, the switch reads the first 64 bytes of the packet before forwarding it on. Thus, it acts as a forward buffering switch with respect to short frames and as a cut-through switching with respect to long frames. Personnel promotion methods are illustrated in Figure 3.

(1x1)

Figure 3.
Packet forwarding mechanisms differ in the point at which the packet is forwarded.

The internal bus architecture determines how frames are transferred from one port to another using the switch's internal electronics. It is critical to the efficiency of the switch: a manufacturer may claim that the internal bus has a throughput of 1-2 Gbps, but at the same time keep silent that it is achieved only with a certain type of traffic. For example, a switch with low-capacity buffers can only perform at its best if all ports are running at the same speed and traffic is distributed evenly across all ports.

The bus can service ports cyclically or by priority. During cyclic maintenance, the idle port is skipped. This architecture is best suited for situations where traffic on each port is approximately the same. In priority servicing, active ports compete with each other for the internal bus. This kind of architecture is best suited when working with switches whose ports have different speed. Some manufacturers offer switches with the ability to change the type of bus architecture.

FULL DUPLEX ETHERNET

Regular Ethernet (and Fast Ethernet) is a shared transmission medium, and all shared networks are half-duplex by definition: at a given time, only one station has the right to transmit, and everyone else must listen to it. Or, to put it another way, a station can perform either receiving or transmitting, but not both at the same time.

The widespread adoption of four-pair wiring has opened up the fundamental possibility of transmitting and receiving data over separate paths (different pairs), which did not exist when the physical transmission medium was coaxial cable.

In the case when only one node is connected to each port of the switch (we emphasize, one), there is no contention for access to the transmission medium, so no collisions can arise in principle and the CSMA/CD multiple access scheme is no longer needed.

Thus, if two nodes are connected directly to the switch ports, then they can receive and transmit data simultaneously on different pairs, resulting in a theoretical throughput of such a connection of 20 Mbit/s in the case of Ethernet and 200 Mbit/s in the case of Fast Ethernet. In addition, due to the absence of competition, the real average connection throughput is close to the nominal one and is over 80% of the above values.

AUTOMATIC NEGOTIATION

Some switches have both 10 Mbps and 100 Mbps ports (see the "Preventing Congestion" section for information on what problems this can cause). Moreover, they are able to automatically determine at what speed stations, hubs, etc. connected to it are operating. Finally, if only one node is connected to a switch port, then both sides can select full-duplex operation mode (provided that it is supported by both ).

The same standard RJ-45 connector can carry 10BaseT, 10BaseT full duplex, 100BaseTX, 100BaseTX full duplex, and 100BaseT4 signals. Therefore, IEEE proposed an automatic mode negotiation scheme called nWAY to determine which standard the device at the other end of the cable is operating on. The order of priority for operating modes is as follows:

full duplex 100BaseTX;
100BaseT4;
100BaseTX;
full duplex 10BaseT;
10BaseT.

In autonegotiation, the “contracting parties” use a 10BaseT analogue of Link Integrity pulses called Fast Link Pulse. Such pulses are sent by both devices, and each of them uses them to determine which transmission mode the other side is capable of operating in.

Many switches support all five possible modes, so even if the connected host does not have auto-negotiation, the switch port will communicate with it on that maximum speed which he is capable of. In addition, the implementation of this function is very simple and does not lead to any noticeable increase in the cost of equipment. Finally, the standard provides the ability to disable auto-negotiation so that the user can set desired mode transmission manually, if necessary.

PREVENTING OVERLOADS

Switches often need to act as a bridge between 10 and 100 Mbps ports, for example, when the switch has one high-speed port for connecting a server and a number of 10 Mbps ports for connecting workstations. In the case when traffic is transmitted from a 10 Mbit/s port to a 100 Mbit/s port, no problems arise, but if the traffic goes in the opposite direction... Data flow of 100 Mbit/s

is an order of magnitude greater than the capabilities of a 10 Mbps port, so the switch must store redundant data in its internal buffers if it has sufficient memory to do so. For example, let's say the first port is connected to a server with a 100 Mbps card, and the second port is connected to a client with a 10 Mbps card. If the server sends 16 packets in a row to the client one after the other, then together they amount to an average of 24 KB of data. Transmitting a 1.5 KB frame takes 122 µs in the case of Fast Ethernet and 1220 µs in the case of Ethernet. Thus, the first port will receive ten frames before one frame can be sent through the second port, i.e., the first port must have a buffer of at least 24 KB. However, if the stream is long enough, then no buffers will be enough. One way to avoid congestion is through thread management. The concept of flow control (or congestion avoidance) involves causing an artificial collision on a high-speed port, as a result of which the sender suspends data transmission for some time in accordance with the exponential fallback algorithm. In our example, the first port will detect that its buffer is full and send a congestion message back to the sender. The latter will perceive this message as a collision and will pause the transmission. The switch will continue to send congestion messages until the buffer becomes free. This kind of flow control is performed only by switches with half-duplex ports.

SWITCH MANAGEMENT

Monitoring switch performance is one of the biggest challenges facing both equipment manufacturers and network administrators. In the case of shared networks, management is not particularly difficult, since traffic through one port is forwarded to all other ports on the hub. In the case of a switch, the traffic between pairs of ports of each virtual connection is different, so the task of collecting statistical data about the operation of the router is much more complicated. Manufacturers generally support the following two methods of collecting statistics.

One is to incorporate management into the switch backplane architecture. Statistics are collected about each packet transmitted over the bus and stored in the control device in accordance with its MAC address. The management program can access this device for statistics over the local network. The only problem with this method is that each switch manufacturer implements its own design, so compatibility is usually limited to SNMP statistics.

The second method is known as port mirroring. In this case, all traffic on the specified port is copied to the dedicated management port. This port usually connects to the control terminal, which already collects statistics for each specific port. However, this method has the limitation that it does not allow you to see what is happening at this time on other ports of the switch.

Some switch manufacturers include in their models, as a rule, high-end remote monitoring information bases (Remote Monitor MIB, RMON) in order to collect statistics on the functioning of each switch port. But very often they do not include all groups defined by the standard, and, in addition, support for RMON MIB significantly increases the cost of the switch.

VARIETIES OF SWITCHES

Switches can be classified in different ways. Based on their purpose, they can all be divided into two large groups - switches for workgroups and switches for the backbone.

A distinctive feature of many workgroup switches is the small number of addresses supported on each port. Every port acts as a bridge, so it must know which addresses it can access through other ports. Such lists of port-to-MAC address mappings can be quite long and take up a significant amount of expensive memory. Therefore, workgroup switches usually do not support too many MAC addresses. Some of them generally remember only one address for each port - in this case, one and only one node can be connected to the port.

Backbone switches are distinguished by a large number of high-speed ports, including full-duplex, and the presence additional functions network management such as virtual local area networks and advanced packet filtering, etc. In general, a backbone switch is much more expensive and more productive than its workgroup counterpart.

ADVANTAGES OF SWITCHING

Switching has become such a popular technology because it allows you to increase the actual bandwidth available to each node. As a result, without changing the underlying technology or significantly redesigning the network topology, companies were able to clear traffic jams and expand bottlenecks. In addition, it allows you to increase the length of the network. This circumstance is especially valuable in the case of Fast Ethernet - for example, by installing a bridge (a two-port switch, from the point of view of some manufacturers) between two hubs, the distance between end stations can be increased to 400 m.

Dmitry Ganzha is the executive editor of LAN. He can be contacted at: .

From shared to switched networks