Building networks using ethernet 1000base t technology. Development of an integrated access network based on Ethernet and Wi-Fi technologies

Ethernet technology template, written in the IEEE 802.3 dock. This is the only description of a MAC layer format frame. In the Ethernet network, only one type of frame of the link layer is implemented, the header of which is a set of headers of the MAC and LLC sublayers, which is some kind.

Ethernet DIX / Ethernet II, appeared in 1980 as a result of joint robots of three firms Xerox, Intel and Digital, which introduced version 802.3 as an international standard;
The committee adopted 802.3 and revised it slightly. This is how 802.3 / LLC, 802.3 / 802.2 or Novell 802.2;
Raw 802.3 or Novell 802.3- designed to speed up the work of their protocol stack in Ethernet networks;
Ethernet SNAP is the result of the 802.2 committee which has been brought to a common standard and has become flexible for future possible addition of fields;

Today, network hardware and software can handle all frame formats, and frame recognition works automatically which reduces and one of the. The frame formats are shown in Figure 1.

Picture 1

802.3 / LLC frame

The header of this frame combines the header fields of the IEEE 802.3 and 802.2 frames. The 802.3 standard consists of:

Preamble field- called the field of sync bytes - 10101010. In Manchester coding, this code is modified in the physical medium into a signal with a frequency of 5 MHz.
Starting frame delimiter- is one byte 10101011. This field indicates that the next byte is the first byte of the frame header.
Destination address- This field can be 6 or 2 bytes long. Typically this field is used for a 6 byte MAC address.
Source address Is a field that contains 6 or 2 bytes of the MAC address of the sender node. The first bit is always - 0.
Length- a field that has a size of 2 bytes, and contains the length of the data field in the frame.
Data field- the field can be from 0 to 1500 bytes. But if suddenly the data occupies less than 46 bytes, then the field is used placeholder which pads the field to 46 bytes.
Placeholder field- Provides filling of the data field, if its weight is less than 46 bytes. Needed for the correct operation of the collision detection mechanism.
Frame check sequence field- this field contains a control sum of 4 bytes. CRC-32 algorithm is used /

This frame is a MAC sublayer frame, its data field contains an LLC sublayer frame with removed flags at the end and beginning of the frame that is transmitted through.

Raw 802.3 / Novell 802.3 frame

This frame used to be a network layer protocol in MetWare OS. But now that the need to identify the upper layer protocol has disappeared, the frame has been encapsulated in the MAC frame of the LLC frame.

Ethernet DIX / Ethernet II frame

This frame has a structure that is similar to that of Ras 802.3. But the 2-byte length field here has protocol type field assignments. Indicates the type of upper layer protocol that nested its packet in the data field of this frame. These frames are distinguished by the length of the field, if the value is less than 1500, then this is the length field, if more, then the type.

Ethernet SNAP frame

The frame appeared as a result of eliminating the inconsistency in the encoding of the protocol types. The protocol is also used in the IP protocol when encapsulating the following networks: Token Ring, FDDI, 100VC-AnyLan. But when transmitting IP packets over Ethernet, the protocol uses Ethernet DIX frames.

IPX protocol

This protocol can use all four Ethernet frame types. It defines the type by checking the absence or presence of the LLC field. Also behind the DSAP / SSAP fields. If the field value is 0xAA, then this is a SNAP frame, otherwise it is 802.3 / LLC.

Data transmitted over Ethernet is split into frames. Recall that almost every network technology (regardless of its level) corresponds to a unit of data transmission: Ethernet - frame, ATM - cell, IP - datagram, etc. Pure data is not transmitted over the network. As a rule, a header is "attached" to the data unit. Some networking technologies also add an ending. The heading and ending carry service information and consist of certain fields.

Since there are several types of frames, in order to understand each other, the sender and receiver must use the same frame type. Frames can be of four different formats, slightly different from each other. There are only two raw formats - Ethernet II and Ethernet 802.3. These formats differ in the purpose of only one field.

For successful delivery of information to the recipient, each frame must, in addition to data, contain service information: the length of the data field, physical addresses of the sender and recipient, the type of network protocol, etc.

In order for workstations to be able to interact with a server on the same network segment, they must support a single frame format. There are four main flavors of Ethernet frames:

Ethernet Type II
Ethernet 802.3
Ethernet 802.2
Ethernet SNAP (SubNetwork Access Protocol).

Consider the fields common to all four types of frames (Fig. 1).

Rice. 1. General Ethernet Frame Format

The fields in the frame have the following meanings:

The "Preamble" and "Start of frame" fields are used to synchronize the sender and the receiver. The preamble is a 7-byte sequence of ones and zeros. The frame start flag field has a size of 1 byte. These fields are not taken into account when calculating the frame length.
The "Destination address" field consists of 6 bytes and contains the physical address of the device on the network to which this frame is addressed. The values for this and the next field are unique. Each manufacturer of Ethernet adapters is assigned the first three bytes of the address, and the remaining three bytes are determined directly by the manufacturer. For example, for 3Com adapters, physical addresses will start with 0020AF. The first bit of the recipient address has special meaning. If it is 0, then this is the address of a specific device (only in this case, the first three bytes are used to identify the manufacturer of the network card), and if 1 - broadcast. Typically, in a broadcast address, all remaining bits are also set to one (FF FF FF FF FF FF).
The field "Sender address" consists of 6 bytes and contains the physical address of the device on the network that sent this frame. The first bit of the sender address is always zero.
The Length / Type field can contain the length or type of the frame, depending on the Ethernet frame in use. If the field specifies a length, it is specified in two bytes. If type - then the content of the field indicates the type of the upper-level protocol to which this frame belongs. For example, for IPX, the field is 8137, and for IP, 0800.
The "Data" field contains the data of the frame. More often than not, this is information needed by upper-level protocols. This field has no fixed length.
The "Checksum" field contains the result of calculating the checksum of all fields, except for the perambula, the start of the frame sign and the checksum itself. The computation is performed by the sender and added to the frame. A similar calculation procedure is performed on the recipient's device. If the result of the calculation does not match the value of this field, it is assumed that an error occurred during transmission. In this case, the frame is considered corrupted and ignored.

It should be noted that the minimum allowable length of all four types of Ethernet frames is 64 bytes, and the maximum is 1518 bytes. Since 18 bytes are allocated for service information in the frame, the "Data" field can have a length from 46 to 1500 bytes. If the data transmitted over the network is less than the allowed minimum length, the frame will be automatically padded to 46 bytes. Such strict restrictions on the minimum frame length were introduced to ensure the normal operation of the collision detection mechanism.

ETHERNET TECHNOLOGY

Ethernet is the most widely used standard for local area networks today.

When they say Ethernet, they usually mean any of the variants of this technology. More specifically, Ethernet is a networking standard based on the experimental Ethernet Network that Xerox developed and implemented in 1975. The access method was tried even earlier: in the second half of the 60s, various options for random access to a common radio environment, collectively called Aloha, were used in the radio network of the University of Hawaii. In 1980, DEC, Intel, and Xerox jointly developed and published the Ethernet version II standard for a coaxial cable network, which was the latest version of the proprietary Ethernet standard. Therefore, the proprietary version of the Ethernet standard is called the Ethernet DIX or Ethernet P.

Based on the Ethernet DIX standard, the IEEE 802.3 standard was developed, which in many respects coincides with its predecessor, but there are still some differences. While the IEEE 802.3 standard differentiates between MAC and LLC layers, the original Ethernet combined both layers into a single data link layer. Ethernet DIX defines an Ethernet Configuration Test Protocol that IEEE 802.3 does not. The frame format is also slightly different, although the minimum and maximum frame sizes in these standards are the same. Often, in order to distinguish Ethernet, as defined by the IEEE standard, and proprietary Ethernet DIX, the former is called 802.3 technology, and the proprietary Ethernet name is left without additional designations.

Depending on the type of physical medium, the IEEE 802.3 standard has various modifications - 10Base-5, 10Base-2, 10Base-T, 10Base-FL, 10Base-FB.

In 1995, the Fast Ethernet standard was adopted, which in many respects is not an independent standard, as evidenced by the fact that its description is simply an additional section to the main 802.3 standard - the 802.3u section. Similarly, the 1998 Gigabit Ethernet standard is described in the 802.3z section of the main document.

For transmission of binary information over cable for all variants of the physical layer of Ethernet technology, providing a throughput of 10 Mbit / s, the Manchester code is used.

All types of Ethernet standards (including Fast Ethernet and Gigabit Ethernet) use the same media separation method - the CSMA / CD method.

Ethernet Addressing

To identify the recipient of information in Ethernet technologies, 6-byte MAC addresses are used.

The MAC address format provides the ability to use specific multicast addressing modes in the Ethernet network and, at the same time, exclude the possibility of two stations that would have the same address within the same local network.

The physical address of an Ethernet network consists of two parts:

Vendor codes
Individual device identifier

A special organization within the IEEE is engaged in the distribution of permitted encodings of this field at the request of manufacturers of network equipment. Various forms can be used to write the MAC address. The most commonly used form is hexadecimal, in which pairs of bytes are separated by "-" characters:

E0-14-00-00-00

In Ethernet and IEEE 802.3 networks, there are three main modes of forming the destination address:

Unicast - individual address;
Multicast - multicast address;
Broadcast - broadcast address.

The first addressing mode (Unicast) is used when the source station addresses the transmitted packet to only one recipient of the data.

A sign of using the Multicast addressing mode is the presence of 1 in the least significant bit of the most significant byte of the equipment manufacturer's identifier.

C-CC-CC-CC

A frame whose DA field content belongs to the Multicast type will be received and processed by all stations that have a corresponding Vendor Code value - in this case, these are Cisco network devices. Given Multicast - the address is used by the network devices of this company to interact in accordance with the rules of the Cisco Discovery Protocol (CDP).

An Ethernet and IEEE 802.3 station can also use Broadcast addressing mode. The address of the Broadcast destination station is encoded with a special value:

FF-FF-FF-FF-FF-FF

When using this address, the transmitted packet will be received by all stations that are in this network.

CSMA / CD Access Method

Ethernet networks use a media access method called carrier-sense-multiply-access with collision detection (CSMA / CD) ...

The CSMA / CD protocol defines the nature of the interaction of workstations in a network with a single common data transmission medium for all devices. All stations have equal conditions for data transmission. There is no specific sequence in which stations can access the medium for transmission. It is in this sense that the environment is accessed randomly. The implementation of random access algorithms seems to be a much simpler task than the implementation of deterministic access algorithms. Since in the latter case, either a special protocol is required that controls the operation of all network devices (for example, the token circulation protocol inherent in Token Ring and FDDI networks), or a special dedicated device - a master hub, which, in a certain sequence, would provide all the rest of the station with the ability to transmit (Arcnet networks , 100VG AnyLAN).

However, the network with random access has one, perhaps, the main drawback - it is not entirely stable network operation under heavy load, when a sufficiently long time can pass before a given station is able to transmit data. This is due to collisions that arise between stations that started transmitting at the same time or almost simultaneously. In the event of a collision, the transmitted data does not reach the recipients, and the transmitting stations have to restart the transmission again - the coding methods used in Ethernet do not allow the signals of each station to be separated from the general signal. (Z Note that this fact is reflected in the “Base (band)” component present in the names of all physical protocols of Ethernet technology (for example, 10Base-2,10Base-T, etc.). Baseband network means a baseband network in which messages are sent digitally over a single channel, without frequency division.)

Collision is a normal situation in Ethernet networks. For a collision to occur, it is not necessary for several stations to start transmitting absolutely simultaneously, such a situation is unlikely. It is much more likely that a collision occurs due to the fact that one node starts transmitting earlier than the other, but the signals of the first simply do not have time to reach the second node by the time the second node decides to start transmitting its frame. That is, collisions are a consequence of the distributed nature of the network.

The set of all stations on the network, the simultaneous transmission of any pair of which leads to a collision, is called a collision domain or collision domain.

Collisions can cause unpredictable delays in the propagation of frames over the network, especially when the network is heavily loaded (many stations are trying to simultaneously transmit within the collision domain,> 20-25) and when the collision domain is large (> 2 km). Therefore, when building networks, it is advisable to avoid such extreme operating modes.

The problem of constructing a protocol capable of resolving collisions in the most optimal way, and optimizing the network operation at high loads, was one of the key issues at the stage of the standard formation. Initially, three main approaches were considered as candidates for the implementation of an algorithm for random access to the environment: non-persistent, 1-constant, and p-constant (Figure 11.2).

Figure 11.2. Multiple random access (CSMA) algorithms and collision back off

Nonpersistent algorithm. With this algorithm, the station wishing to transmit is guided by the following rules.

1. Listens to the medium, and if the medium is free (ie, if there is no other transmission or there is no collision signal) it transmits, otherwise - the medium is busy - go to step 2;

2. If the environment is busy, it waits for a random (in accordance with a certain probability distribution curve) time and returns to step 1.

Using a random wait value in a busy environment reduces the likelihood of collisions. Indeed, suppose otherwise that two stations are going to transmit almost simultaneously, while the third is already transmitting. If the first two would not have a random waiting time before the start of the transmission (in case the environment turned out to be busy), but only listened to the environment and waited for it to become free, then after the third station stopped transmitting, the first two would start transmitting simultaneously, which would inevitably lead to collisions. Thus, random waiting eliminates the possibility of such collisions. However, the inconvenience of this method is manifested in the inefficient use of the channel bandwidth. Since it may happen that by the time the medium becomes free, the station wishing to transmit will continue to wait for some random time before deciding to listen to the medium, since it had already listened to the medium, which turned out to be busy. As a result, the channel will be idle for a while, even if only one station is waiting for transmission.

1-persistent algorithm... To reduce the time when the environment is not busy, a 1-persistent algorithm could be used. With this algorithm, the station wishing to transmit is guided by the following rules.

1. Listens to the medium, and if the medium is idle, transmits, otherwise, go to step 2;

2. If the medium is busy, it continues to listen to the medium until the medium is free, and as soon as the medium is released, it immediately starts transmitting.

Comparing the non-persistent and 1-persistent algorithms, we can say that in the 1-persistent algorithm the station wishing to transmit behaves more "selfishly". Therefore, if two or more stations are waiting for transmission (waiting until the environment is free), a collision, one might say, will be guaranteed. After the collision, the stations begin to think about what to do next.

P-persistent algorithm. The rules for this algorithm are as follows:

1. If the environment is free, the station with the probability p immediately starts transmission or with probability (1- p ) waits for a fixed time interval T. The interval T is usually taken equal to the maximum propagation time of the signal from end to end;

2. If the medium is busy, the station continues listening until the medium is free, then goes to step 1;

3. If the transmission is delayed by one interval T, the station returns to step 1.

And here the question arises of choosing the most effective value of the parameter p ... The main problem is how to avoid instability at high loads. Consider a situation in which n stations intend to transmit frames while transmission is already in progress. At the end of the transmission, the expected number of stations that will try to transmit will be equal to the product of the number of stations willing to transmit by the transmission probability, that is np ... If np > 1, then on average several stations will try to transmit at once, which will cause a collision. Moreover, as soon as a collision is detected, all stations will go back to step 1, which will cause a second collision. In the worst case, new stations willing to betray may be added to n , which will further exacerbate the situation, leading eventually to continuous collision and zero throughput. To avoid such a disaster, the work np should be less than one. If the network is susceptible to the occurrence of conditions when many stations simultaneously wish to transmit, then it is necessary to reduce p ... On the other hand, when p become too small, even a single station can wait on average (1- p )/p intervals T before transmitting. So if p = 0.1 then the average idle time prior to the transfer will be 9T.

CSMA / CD Collision Resolution Multiple Medium Access Protocol embodied the ideas of the above algorithms and added an important element - collision resolution. Since a collision destroys all frames transmitted at the moment of its formation, then there is no point in stations to continue further transmission of their frames, as soon as they (stations) have detected collisions. Otherwise, there would be a significant loss of time when transmitting long frames. Therefore, for timely detection of collisions, the station listens to the environment throughout its own transmission. Here are the basic rules of the CSMA / CD algorithm for the transmitting station (Figure 11.3):

1. The station about to transmit is listening to the environment. And transmits if the environment is free. Otherwise (that is, if the environment is busy) proceeds to step 2. When transmitting several frames in a row, the station maintains a certain pause between frame transmission - the interframe interval, and after each such pause before sending the next frame, the station again listens to the environment (return to the beginning step 1);

2. If the environment is busy, the station continues to listen on the environment until the environment becomes free, and then immediately starts transmitting;

3. Each station transmitting listens to the environment, and if a collision is detected, it does not immediately stop transmitting, but first transmits a short special collision signal - a jam-signal, informing other stations about the collision, and stops transmitting;

4. After transmitting the jam-signal, the station stops talking and waits for some arbitrary time in accordance with the rule of binary exponential delay and then returns to step 1.

To be able to transmit a frame, the station must ensure that the shared medium is free. This is accomplished by listening to the fundamental of the signal, also called carrier-sense (CS). A sign of an unoccupied environment is the absence of a carrier frequency on it, which with the Manchester coding method is equal to 5-10 MHz, depending on the sequence of ones and zeros transmitted at the moment.

After the end of the frame transmission, all network nodes must withstand a technological pause (Inter Packet Gap) of 9.6 μs (96 bt). This pause, also called interframe spacing, is needed to bring the network adapters to their original state and also to prevent a single station from taking over the media exclusively.

Figure 11.3. Block diagram of the CSMA / CD algorithm (MAC level): when transmitting a frame by a station

Jam signal (jamming - literally jamming). The transmission of a jam signal guarantees that more than one frame will not be lost, since all nodes that transmitted frames before the collision occurred, having received a jam signal, will interrupt their transmissions and become silent in anticipation of a new attempt to transmit frames. The Jam signal must be of sufficient length to reach the most distant stations in the collision domain, taking into account the additional safety margin (SF) delay on possible repeaters. The content of the jam signal is not critical, except that it should not match the CRC field of the partially transmitted frame (802.3), and the first 62 bits should represent an alternation of ‘1’ and ‘0’ with a start bit ‘1’.

Figure 11.4. Random access method CSMA / CD

Figure 11.5 illustrates the collision detection process for a bus topology (thin or thick coaxial cable (10Base5 and 10Base2, respectively).

At the moment of time the node A(DTE A) starts transmission, naturally listening to its own transmitted signal. At the moment in time when the frame has almost reached the node B(DTE B), this node, not knowing that a transmission is already in progress, starts transmitting itself. At a point in time, a node B detects a collision (the constant component of the electrical signal in the monitored line increases). After that the node B transmits a jam signal and stops transmission. At the moment of time, the collision signal reaches the node A, then A also transmits a jam signal and stops transmission.

Figure 11.5. Collision detection when using the CSMA / CD scheme

According to the IEEE 802.3 standard, a node cannot transmit very short frames, or in other words, conduct very short transmissions. Even if the data field is not filled to the end, a special additional field appears that extends the frame to a minimum length of 64 bytes, excluding the preamble. Channel time ST (slot time) is the minimum time during which a node is obliged to transmit, to occupy a channel. This time corresponds to the transmission of a frame of the minimum allowable size, accepted by the standard. Channel time is related to the maximum allowable distance between network nodes - the diameter of the collision domain. Lets say the above example implements a worst-case scenario where the stations A and B removed from each other at the maximum distance. Time, signal propagation from A before B denote by. Knot A starts transmitting at time zero. Knot B starts transmitting at a moment in time and detects a collision after an interval after the start of its transmission. Knot A detects a collision at a point in time. In order for the frame emitted A, was not lost, it is necessary that the node A did not stop transmitting to this moment, since then, having detected a collision, the node A will know that his frame has not arrived and will try to transmit it again. Otherwise, the frame will be lost. The maximum time after which from the moment of the beginning of the transfer the node A can still detect a collision equals - this time is called double turnover time PDV (Path Delay Value, PDV)... More generally, PDV defines the total delay associated both with the delay due to the finite segment length and with the delay arising in the processing of frames at the physical layer of intermediate repeaters and end nodes of the network. For further consideration, it is also convenient to use another unit of time measurement: bit time bt (bit time). The time of 1 bt corresponds to the time it takes to transmit one bit, i.e. 0.1 μs at 10 Mbps.

Clear recognition of collisions by all stations on the network is a prerequisite for the correct operation of the Ethernet network. If any transmitting station does not recognize the collision and decides that the data frame was transmitted by it correctly, then this data frame will be lost. Due to the overlap of signals during a collision, the information of the frame will be distorted, and it will be rejected by the receiving station (possibly due to a mismatch of the checksum). Most likely, the garbled information will be retransmitted by some higher-level protocol, such as a transport or connection-oriented application protocol. But the retransmission of the message by the upper-layer protocols will occur at a much longer time interval (sometimes even several seconds) compared to the microsecond intervals that the Ethernet protocol operates. Therefore, if collisions are not reliably recognized by the nodes of the Ethernet network, this will lead to a noticeable decrease in the useful bandwidth of this network.

For reliable collision detection, the following relationship must be met:

T min> = PVD,

where T min is the transmission time of the minimum frame length, and PDV is the time it takes for the collision signal to propagate to the farthest network node. Since in the worst case, the signal must pass twice between the stations of the network most distant from each other (an undistorted signal passes in one direction, and on the way back, the signal already distorted by the collision propagates), that is why this time is called double turnover time (Path Delay Value, PDV).

When this condition is met, the transmitting station must have time to detect the collision caused by its transmitted frame, even before it finishes transmitting this frame.

Obviously, the fulfillment of this condition depends, on the one hand, on the length of the minimum frame and the network bandwidth, and on the other hand, on the length of the network cable system and the speed of signal propagation in the cable (for different types of cable, this speed is somewhat different).

All parameters of the Ethernet protocol are selected in such a way that during normal operation of network nodes, collisions are always clearly recognized. When choosing the parameters, of course, the above relationship was taken into account, linking the minimum frame length and the maximum distance between stations in the network segment.

In the Ethernet standard, it is accepted that the minimum length of the frame data field is 46 bytes (which, together with the service fields, gives the minimum frame length of 64 bytes, and together with the preamble - 72 bytes or 576 bits).

When transmitting large frames, for example 1500 bytes, a collision, if it occurs at all, is detected almost at the very beginning of the transmission, no later than the first 64 transmitted bytes (if a collision did not occur at this time, then later it will not arise, since all stations are listening to the line and, "hearing" the transmission, they will be silent). Since the jam-signal is much shorter than the full frame size, then when using the CSMA / CD algorithm, the amount in the idle used channel capacity is reduced to the time required for collision detection. Early collision detection leads to more efficient channel utilization. Late collision detection, inherent in more extended networks, when the collision domain is several kilometers in diameter, which reduces the efficiency of the network. Based on a simplified theoretical model of the behavior of a busy network (assuming a large number of simultaneously transmitting stations and a fixed minimum length of transmitted frames for all stations), it is possible to express the network performance U in terms of the PDV / ST ratio:

where is the base of the natural logarithm. Network performance is affected by the size of the frames being broadcast and the diameter of the network. Performance in the worst case (when PDV = ST) is about 37%, and in the best case (when PDV is much less than ST) tends to 1. Although the formula is derived in the limit of a large number of stations trying to transmit simultaneously, it does not take into account the peculiarities of the truncated binary exponential delay algorithm, considered below, and is not valid for a network heavily congested with collisions, for example, when there are more than 15 stations wishing to transmit.

Truncated binary exponential delay(truncated binary exponential backoff). The CSMA / CD algorithm, adopted in the IEEE 802.3 standard, is the closest to the 1-constant algorithm, but it has an additional element - a truncated binary exponential delay. When a collision occurs, the station counts the number of times a collision occurs in a row when sending a packet. Since repeated collisions indicate a high load on the environment, the MAC tries to increase the delay between retrying frame transmissions. The corresponding procedure for increasing time intervals obeys the rule truncated binary exponential delay.

A random pause is selected according to the following algorithm:

Pause = Lx (delay interval),

where (backoff interval) = 512 bit intervals (51.2 μs);

L is an integer selected with equal probability from the range, where N is the retry number of the given frame: 1,2, ..., 10.

After the 10th attempt, the interval from which the pause is selected does not increase. Thus, a random pause can range from 0 to 52.4 ms.

If 16 consecutive attempts to transmit a frame cause a collision, then the transmitter should stop trying and discard this frame.

The CSMA / CD algorithm using truncated binary exponential latency is recognized as the best among the many random access algorithms and provides efficient network operation at both low and medium loads. At high loads, two disadvantages should be noted. First, with a large number of collisions, station 1, which is about to send a frame for the first time (before that has not tried to transmit frames), has an advantage over station 2, which has already tried unsuccessfully to transmit a frame several times, encountering collisions. Because station 2 waits for a significant amount of time before subsequent attempts, according to the binary exponential delay rule. Thus, irregular frame transmission can occur, which is undesirable for time-dependent applications. Secondly, under heavy workload, the efficiency of the network as a whole decreases. Estimates show that with the simultaneous transmission of 25 stations, the total bandwidth is reduced by about 2 times. But the number of stations in the collision domain can be larger, since not all of them will simultaneously access the environment.

Receiving a frame (fig. 11.6)

Figure 11.6. Block diagram of the CSMA / CD algorithm (MAC level): when a frame is received by a station

The receiving station or other network device, for example, a hub or switch, first synchronizes with the preamble and then converts the Manchester code into binary form (at the physical layer). Next, the binary stream is processed.

At the MAC level, the remaining preamble bits are cleared and the station reads the destination address and compares it to its own. If the addresses match, then the frame fields except for the preamble, SDF and FCS are buffered and a checksum is calculated, which is compared with the check sequence field of the FCS frame (using the CRC-32 cyclic sum method). If they are equal, then the contents of the buffer are passed to the higher layer protocol. Otherwise, the frame is discarded. The occurrence of a collision when receiving a frame is detected either by a change in the electrical potential if a coaxial segment is used, or by the fact of receiving a defective frame, an incorrect checksum if a twisted pair or optical fiber is used. In both cases, the received information is discarded.

From the description of the access method, it can be seen that it is probabilistic in nature, and the probability of successfully obtaining a common environment at its disposal depends on the network congestion, that is, on the intensity of the need for frame transmission in the stations. When developing this method in the late 70s, it was assumed that the data transfer rate of 10 Mbps is very high compared to the needs of computers for mutual data exchange, so the network load will always be small. This assumption is sometimes true to this day, but there are already real-time multimedia applications that are very busy on Ethernet segments. In this case, collisions occur much more often. With significant collision rates, the usable throughput of the Ethernet network drops sharply, as the network is almost constantly busy with retrying frame transmissions. To reduce the intensity of collisions, you need to either reduce traffic by reducing, for example, the number of nodes in a segment or replacing applications, or to increase the speed of the protocol, for example, switch to Fast Ethernet.

It should be noted that the CSMA / CD access method does not at all guarantee a station that it will ever be able to access the medium. Of course, with a low network load, the probability of such an event is small, but with a network utilization rate approaching 1, such an event becomes very likely. This shortcoming of the random access method is a price to pay for its extreme simplicity, which made Ethernet the least expensive technology. Other access methods - Token Ring and FDDI token access, the Demand Priority method of 100VG-AnyLAN networks - are free from this drawback.

As a result of taking into account all factors, the ratio between the minimum frame length and the maximum possible distance between network stations was carefully selected, which ensures reliable collision detection. This distance is also called the maximum network diameter.

With an increase in the frame rate, which takes place in new standards based on the same CSMA / CD access method, for example Fast Ethernet, the maximum distance between network stations decreases in proportion to the increase in the transfer rate. In the Fast Ethernet standard, it is about 210 meters, and in the Gigabit Ethernet standard, it would be limited to 25 meters, if the developers of the standard did not take some measures to increase the minimum packet size.

Table 11.1 shows the values of the basic parameters of the 802.3 standard frame transmission procedure, which do not depend on the implementation of the physical medium. It is important to note that each variant of the physical environment of Ethernet technology adds to these constraints its own, often more stringent constraints, which must also be met and which will be discussed below.

Table 11.1.Ethernet MAC Layer Parameters

Parameters	The values
Bit rate	10 Mbps
Grace period	512 bt
Interframe Gap (IPG)	9.6 μs
Maximum number of transmission attempts
Maximum number of increasing pause range
Jam sequence length	32 bit
Maximum frame length (without preamble)	1518 bytes
Minimum frame length (without preamble)	64 bytes (512 bits)
Preamble length	64 bit
Minimum length of random pause after collision	0 bt
Maximum length of random pause after collision	524000 bt
Maximum distance between network stations	2500m
Maximum number of stations in the network

Ethernet frame formats

The Ethernet technology standard described in the IEEE 802.3 document describes a single MAC layer frame format. Since the MAC layer frame must include the LLC layer frame described in the IEEE 802.2 document, according to the IEEE standards, only one version of the link layer frame can be used in the Ethernet network, the header of which is a combination of the MAC and LLC sublayer headers.

Nevertheless, in practice in Ethernet networks at the link layer, frames of 4 different formats (types) are used. This is due to the long history of the development of Ethernet technology, which existed before the adoption of the IEEE 802 standards, when the LLC sublayer was not separated from the general protocol and, accordingly, the LLC header was not used.

A consortium of three firms Digital, Intel and Xerox in 1980 submitted to the 802.3 committee their proprietary version of the Ethernet standard (which, of course, described a certain frame format) as a draft international standard, but the 802.3 committee adopted a standard that differs in some details from DIX offers. The differences were also in the frame format, which gave rise to the existence of two different types of frames in Ethernet networks.

Another frame format emerged as a result of Novell's efforts to speed up its protocol stack over Ethernet.

And finally, the fourth frame format is the result of the work of the 802: 2 committee to bring previous frame formats to some common standard.

Differences in frame formats can result in incompatibility between hardware and network software that is designed to work with only one Ethernet frame standard. However, today almost all network adapters, network adapter drivers, bridges / switches and routers can work with all Ethernet technology frame formats used in practice, and the frame type is recognized automatically.

Below is a description of all four types of Ethernet frames (here, a frame means the entire set of fields that are related to the link layer, that is, the fields of the MAC and LLC layers). One and the same frame type can have different names, so below for each frame type are given several of the most common names:

802.3 / LLC frame (802.3 / 802.2 frame or Novell 802.2 frame);
Raw 802.3 frame (or Novell 802.3 frame);
Ethernet DIX frame (or Ethernet II frame);
Ethernet SNAP frame.

The formats for all of these four types of Ethernet frames are shown in Fig. 11.7.

802.3 / LLC frame

The 802.3 / LLC frame header is the result of the concatenation of the frame header fields defined in the IEEE 802.3 and 802.2 standards.

The 802.3 standard defines eight header fields (Figure 11.7; preamble field and start frame delimiter not shown in the figure).

Preamble field consists of seven sync bytes 10101010. In Manchester coding, this combination is represented in the physical environment by a periodic waveform with a frequency of 5 MHz.
Start-of-frame-delimiter (SFD) consists of one byte 10101011. The occurrence of this bit pattern is an indication that the next byte is the first byte of the frame header.
Destination Address (DA) can be 2 or 6 bytes long. In practice, 6 byte addresses are always used. The first bit of the most significant byte of the destination address is an indication of whether the address is individual or group. If it is 0, then the address is individual (unicast), a if 1, then this multicast address. If the address consists of all ones, that is, it has a hexadecimal representation of 0xFFFFFFFFFFFF, then it is intended for all nodes on the network and is called broadcast address.

In the IEEE Ethernet standards, the least significant bit of a byte is displayed in the leftmost position of the field, and the most significant bit in the rightmost position. This non-standard way of displaying the order of bits in a byte corresponds to the order in which bits are transmitted on the communication line by the Ethernet transmitter. The standards of other organizations, for example RFC IETF, ITU-T, ISO, use the traditional byte representation, where the least significant bit is considered the rightmost bit of the byte, and the most significant bit is the leftmost one. However, the byte order remains traditional. Therefore, when reading the standards published by these organizations, as well as reading data displayed on the screen by the operating system or protocol analyzer, the values of each byte of the Ethernet frame must be mirrored in order to get a correct representation of the meaning of the bits of this byte in accordance with the IEEE documents. For example, a multicast address in IEEE notation of the form 1000 0000 0000 0000 1010 0111 1111 0000 0000 0000 0000 0000 or in hexadecimal notation 80-00-A7-FO-00-00 will most likely be displayed by the protocol analyzer in the traditional form as 01-00-5E-0F-00-00.

Source Address (SA) - it is a 2 or 6 byte field containing the address of the node that sent the frame. The first bit of the address is always 0.
Length (Length, L) - 2-byte field that defines the length of the data field in the frame.
Data field can contain from 0 to 1500 bytes. But if the length of the field is less than 46 bytes, then the next field - the padding field - is used to pad the frame to the minimum allowable value of 46 bytes.
Padding consists of as many padding bytes as possible to provide a minimum data field length of 46 bytes. This ensures that the collision detection mechanism works correctly. If the length of the data field is sufficient, then the padding field does not appear in the frame.
Frame Check Sequence (PCS) consists of 4 bytes containing the checksum. This value is calculated using the CRC-32 algorithm. After receiving a frame, the workstation performs its own checksum calculation for this frame, compares the received value with the value of the checksum field, and thus determines whether the received frame is corrupted.

The 802.3 frame is a MAC sublayer frame, therefore, in accordance with the 802.2 standard, an LLC sublayer frame is embedded in its data field with the start and end flags removed. The LLC frame format has been described above. Since the LLC frame has a header length of 3 (in LLC1 mode) or 4 bytes (in LLC2 mode), the maximum data field size is reduced to 1497 or 1496 bytes.

Figure 11.7. Ethernet frame formats

Similar information.

EtherNet standard IEEE 802.3

It is the most widely used networking technology standard today.

Peculiarities:

works with coaxial cable, twisted pair, optical cables;
topology - bus, star;
access method - CSMA / CD.

The architecture of Ethernet networking technology actually brings together a set of standards that have both common features and differences.

Ethernet technology was developed in conjunction with many of the early projects of Xerox PARC Corporation. It is generally accepted that Ethernet was invented on May 22, 1973, when Robert Metcalfe wrote a memo for the head of PARC on the potential of Ethernet technology. But Metcalfe acquired the legal right to the technology a few years later. In 1976, he and his assistant David Boggs published a brochure entitled Ethernet: Distributed Packet Switching For Local Computer Networks. Metcalfe left Xerox in 1979 and founded 3Com to promote computers and local area networks. He managed to convince DEC, Intel and Xerox to work together and develop the Ethernet standard (DIX). This standard was first published September 30, 1980.

Further development of EtherNet technology:

1982-1993 development of 10Mbps EtherNet;
1995-1998 Fast EtherNet development;
1998-2002 development of GigaBit EtherNet;
2003-2007 development of 10GigaBit EtherNet;
2007-2010 development of 40 and 100GigaBit EtherNet;
2010 to date Terabit Ethernet development.

At the MAC layer, which provides access to the medium and the transmission of the frame, unique 6-byte addresses, called MAC addresses, regulated by the standard, are used to identify the network interfaces of network nodes. Typically, the MAC address is written as six pairs of hexadecimal digits separated by dashes or colons, such as 00-29-5E-3C-5B-88. Each network adapter has a MAC address.

Ethernet MAC address structure:

the first bit of the destination MAC address is called the I / G (individual / group or broadcast) bit. In the source address, it is called the Source Route Indicator;
the second bit determines how the address is assigned;
The three most significant bytes of the address are called the Burned In Address (BIA) or Organizationally UniqueIdentifier (OUI);
the manufacturer is responsible for the uniqueness of the lower three bytes of the address.

Some network programs, in particular wireshark, can immediately display the name of the manufacturer of the given network card instead of the manufacturer code.

EtherNet technology frame format

There are 4 types of frames (frames) in Ethernet networks:

802.3 / LLC frame (or Novell 802.2 frame),
Raw 802.3 frame (or Novell 802.3 frame)
Ethernet DIX frame (or Ethernet II frame),
Ethernet SNAP frame.

In practice, EtherNet equipment uses only one frame format, namely the EtherNet DIX frame, sometimes referred to as the latest DIX frame number.

The first two header fields are for addresses:

DA (Destination Address) - MAC address of the destination node;
SA (Source Address) - MAC address of the sender node. To deliver a frame, one address is enough - the destination address, the source address is placed in the frame so that the host that received the frame knows from whom the frame came and who needs to respond to it.

The T (Type) field contains the conditional code of the upper layer protocol, the data of which is in the data field of the frame, for example, the hexadecimal value 08-00 corresponds to an IP puncture. This field is required to support the interface functions of multiplexing and demultiplexing frames when interworking with upper layer protocols.
Data field. If the length of user data is less than 46 bytes, then this field is padded to the minimum size with padding bytes.
The Frame Check Sequence (FCS) field consists of a 4-byte checksum. This value is calculated using the CRC-32 algorithm.

The EtherNet DIX (II) frame does not reflect the division of the EtherNet link layer into the MAC layer and the LLC layer: its fields support the functions of both layers, for example, the interface functions of the T field belong to the functions of the LLC layer, while all other fields support the functions of the MAC layer.

Consider the EtherNet II frame format using the example of an intercepted packet using a Wireshark network analyzer

Please note that since the MAC address consists of a manufacturer code and an interface number, the network analyzer immediately converts the manufacturer code into the manufacturer's name.

Thus, in EtherNet technology, the MAC addresses act as the destination and destination addresses.

Ethernet technology standards

Physical specifications for Ethernet technology include the following transmission media.

l0Base-5 - coaxial cable with a diameter of 0.5 "(1dm = 2.54cm), called" thick "coaxial cable, with a characteristic impedance of 50Ω.
l0Base-2 - Coaxial cable with a diameter of 0.25 ", called" thin "coaxial cable, with a characteristic impedance of 50Ω.
l0Base-T is an Unshielded Twisted Pair (UTP) cable, category 3,4,5.
l0Base-F - fiber optic cable.

The number 10 denotes the nominal bit rate of the standard data, that is, 10Mbit / s, and the word “Base” is the transmission method at one base frequency. The last character indicates the type of cable.

The cable is used as a mono channel for all stations, the maximum segment length is 500m. The station is connected to the cable through a transceiver - transceiver. The transceiver is connected to the DB-15 connector with an AUI interface cable. Terminators are required at each end to absorb signals propagating through the cable.

Rules "5-4-3" for coaxial networks:

The standard for coaxial cable networks allows the use of no more than 4 repeaters in the network and, accordingly, no more than 5 cable segments. With a maximum cable segment length of 500 m, this gives a maximum network length of 500 * 5 = 2500 m. Only 3 out of 5 segments can be loaded, that is, those to which end nodes are connected. There must be unloaded segments between loaded segments.

l0Base-2

The cable is used as a mono channel for all stations, the maximum segment length is 185 m. To connect the cable to the network card, you need a T-connector, and the cable must have a BNC connector.

The 5-4-3 rule is also used.

l0Base-T

It forms a star-shaped topology based on a hub, the hub acts as a repeater and forms a single mono-channel, the maximum segment length is 100m. The end nodes are connected using two twisted pairs. One pair for transferring data from node to hub is Tx, and the other for transferring data from hub to node is Rx.
4-Hub Rules for Twisted Pair Networks:
The twisted pair standard defines the maximum number of hubs between any two stations on the network, namely 4. This rule is called the "4-hub rule". Obviously, if there should not be more than 4 repeaters between any two network nodes, then the maximum diameter of a twisted pair network is 5 * 100 = 500 m (maximum segment length 100 m).

10Base-F

Functionally, an Ethernet over an optical cable consists of the same elements as a 10Base-T network

The FOIRL (Fiber Optic Inter-Repeater Link) standard is the committee's first 802.3 standard for the use of fiber over Ethernet. Max segment length 1000m, max number of hubs 4, with a total network length of not more than 2500 m.

10Base-FL is a minor improvement on the FOIRL standard. Max segment length 2000 m. The maximum number of hubs is 4, and the maximum network length is 2500 m.

The 10Base-FB standard is intended only for connecting repeaters. End nodes cannot use this standard to connect to hub ports. Max number of hubs 5, max length of one segment 2000 m and maximum network length 2740 m.

Table. Ethernet Physical Layer Specification Parameters

When considering the rule "5-4-3" or "4-hubs", if an imaginary signal from a device of the "switch" type appears on the path of propagation through the cables, the calculation of topological constraints starts from zero.

Ethernet bandwidth

Bandwidth is measured in terms of the number of frames or the number of bytes of data transmitted over the network per unit of time. If no collisions occur on the network, the maximum frame rate for the smallest frame size (64 bytes) is 14881 frames per second. At the same time, the useful bandwidth for Ethernet II frames is 5.48 Mbps.

The maximum frame rate for the maximum frame size (1500 bytes) is 813 frames per second. The useful bandwidth will be 9.76 Mbps.

The Preamble (7 bytes) and Initial Frame Delimiter (SFD) (1 byte) frame fields in Ethernet are used for synchronization between the sending and receiving devices. These first eight bytes of the frame are used to draw the attention of the receiving nodes. Essentially, the first few bytes tell receivers to prepare to receive a new frame.

Destination MAC address field

The Destination MAC Address field (6 bytes) is the identifier for the intended recipient. As you may recall, this address is used by Layer 2 to help devices determine if a given frame is being addressed to them. The address in the frame is compared with the MAC address of the device. If the addresses match, the device receives the frame.

Source MAC address field

The Destination MAC Address field (6 bytes) identifies the sending NIC or frame interface. Switches also use this address to add it to their mapping tables. The role of switches will be discussed later in this section.

Field Length / Type

For any IEEE 802.3 standard earlier than 1997, the Length field specifies the exact length of the frame data field. This is later used later as part of the FCS to ensure that the message was received correctly. If the purpose of the field is to define a type, as in Ethernet II, the Type field describes which protocol is being implemented.

These two uses of the field were officially combined in 1997 in the IEEE 802.3x standard because both applications were common. The Ethernet II Type field is included in the current 802.3 frame definition. When a node receives a frame, it must examine the Length field to determine which higher layer protocol is present in it. If the value of two octets is greater than or equal to hexadecimal 0x0600 or decimal 1536, then the contents of the Data field are decoded according to the designated protocol type. If the field value is less than or equal to 0x05DC hexadecimal or 1500 decimal, the Length field is used to indicate the use of the IEEE 802.3 frame format. This differentiates Ethernet II and 802.3 frames.

Fields Data and Padding

The Data and Padding fields (46 - 1500 bytes) contain encapsulated data from a higher layer, which is a typical Layer 3 PDU, usually an IPv4 packet. All frames must be at least 64 bytes long. If a smaller packet is encapsulated, Padding is used to increase the frame size to this minimum size.

The IEEE maintains a list of general purpose Ethernet II types.