Chapter 2: Ethernet, Token Ring, and ATM Frame and Cell Operations
The key to understanding local area network (LAN) performance issues is a detailed level of knowledge concerning the flow of data on a network. To understand how data flows on a LAN we must examine the method by which information is carried within a frame or cell and how different fields are used to provide such functions as routing, error detection, and other network functions. Thus, the purpose of this chapter is to obtain a detailed understanding of the composition of Ethernet and Token Ring frames and ATM cells , including their frame and cell fields and the function of each field within a frame or cell. This information will provide us with the ability to understand the overhead associated with different network frames and cells with respect to the field actually used to transport information. In addition, this information provides a review concerning how network access occurs, network addressing considerations, and similar information we must consider in attempting to determine problems that may occur when establishing or expanding a network or attempting to connect two previously independent networks. In examining LAN frame operations, we first focus attention on the frame format used on an Ethernet network and then examine the three types of transmission formats supported by a Token Ring network. Once the preceding is accomplished, we turn our attention to ATM, examining the composition of the ATM cell, obtaining an overview of ATM connectivity, and computing the overhead associated with the use of a cell-based technology.
2.1 Ethernet Frame Operations
In this section we first look at the composition of different types of Ethernet frames. In actuality, there is only one Ethernet frame, whereas the CSMA/CD (Carrier Sense Multiple Access with Collision Detection) frame format standardized by the IEEE is technically referred to as an 802.3 frame. However, in this book we collectively reference CSMA/CD operations as Ethernet and, when appropriate, indicate differences between Ethernet and the IEEE 802.3 Ethernet-based CSMA/CD standard by a comparison of the two. One such area worthy of a comparison is the frame format, which differs between Ethernet and the IEEE 802.3 Ethernet-based CSMA/CD standard. Once we obtain an understanding of the composition of Ethernet and IEEE 802.3 frames, we will examine the function of fields within each frame and discuss the overhead of the frame with respect to its information transfer capability.
2.1.1 Frame Composition
Figure 2.1 illustrates the general frame composition of Ethernet and IEEE 802.3 frames. A third type of frame that I would be remiss if I did not mention is the Fast Ethernet, 100BASE-TX frame. That frame differs from the IEEE 802.3 frame through the addition of a byte at each end to mark the beginning and end of the frame. Because those bytes do not alter the composition of the frame, I first focus attention on the fields within Ethernet and IEEE 802.3 frames and then describe the bytes unique to Fast Ethernet. Once the preceding is accomplished, I will discuss two techniques used by Gigabit Ethernet to enhance transmission on that network/carrier extension and frame bursting.
Figure 2.1: Ethernet and IEEE 802.3 Frame Formats
In comparing the format of Ethernet and IEEE 802.3 frames, you will note that they slightly differ . An Ethernet frame contains an eight-byte preamble, while the IEEE 802.3 frame contains a seven-byte preamble followed by a one-byte start of frame delimiter field. A second difference between the composition of Ethernet and IEEE 802.3 frames concerns the two-byte Ethernet type field. That field is used by Ethernet to specify the protocol carried in the frame, enabling several protocols to be carried independently of one another. Under the IEEE 802.3 frame format, the type field was replaced by a two-byte length field that specifies the number of bytes that follow that field as data.
The differences between Ethernet and IEEE 802.3 frames, although minor, make the two technically incompatible with one another. Originally, this meant that a network must contain either all Ethernet-compatible network interface cards (NICs) or all IEEE 802.3-compatible NICs. During the late 1980s and early 1990s, software was developed that enabled multiple protocols to be simultaneously transmitted on a LAN, to include different versions of Ethernet. Such software represents multi-protocol drivers, which enables different types of frames to be supported by a common network adapter. Examples of multi-protocol drivers include the Open Data Interface (ODI) and the Network Data Interface Specification (NDIS). Today, the fact that the IEEE 802.3 frame format represents a standard resulted in most vendors now marketing 802.3-compliant hardware and software. Although a few vendors continue to manufacture Ethernet or dual functioning Ethernet/IEEE 802.3 hardware, such products are primarily used to provide organizations with the ability to expand previously developed networks without requiring the wholesale replacement of NICs. Although the IEEE 802.3 standard has essentially replaced Ethernet due to their similarities and the fact that 802.3 was based on Ethernet, we will consider both to be Ethernet. Now that we have an overview of the structure of Ethernet and 802.3 frames, we can probe deeper and examine the composition of each frame field. In doing so we will take advantage of the similarity between Ethernet and IEEE 802.3 frames and examine the fields of each frame on a composite basis, noting the differences between the two when appropriate.
2.1.2 Preamble Field
The preamble field consists of eight (Ethernet) or seven (IEEE 802.3) bytes of alternating 1 and 0 bits. The purpose of this field is to announce the frame as well as enable all receivers on the network to synchronize themselves to the incoming frame. In addition, this field by itself under Ethernet or in conjunction with the start of frame delimiter field under the IEEE 802.3 standard ensures there is a minimum spacing period of 9.6 milliseconds (ms) between frames for error detection and recovery operations. As the speed of Ethernet increases to 100 Mbps or to a Gigabit data rate, the minimum spacing correspondingly decreases. That is, at a 100 Mbps operating rate, the minimum spacing is 0.96 ms between frames, while at a 1 Gbps data rate the minimum spacing is reduced to 0.096 ms.
2.1.3 Start of Frame Delimiter Field
The start of frame delimiter field is only applicable to the IEEE 802.3 standard and can be viewed as a continuation of the preamble. In fact, the composition of this field continues in the same manner as the format of the preamble, with alternating 1 and 0 bits used for the first six bit positions of this one-byte field. The last two bit positions of this field are 11, which breaks the synchronization pattern and alerts the receiver that frame data follows .
Both the preamble field and the start of frame delimiter field are removed by the controller when it places a received frame in its buffer. Similarly, when a controller transmits a frame, it prefixes the frame with those two fields if it is transmitting an IEEE 802.3 frame or a preamble field if it is transmitting a true Ethernet frame.
The removal and addition of the preamble and start of delimiter fields results in a bit of confusion concerning the minimum and maximum length of an Ethernet frame. In some books and trade literature, the minimum and maximum length of Ethernet frames are reduced by eight bytes to reflect the fact that when manipulated within equipment there are no preamble and start of frame delimiter fields. Other literature references minimum and maximum length Ethernet frames to include the preamble and start of frame delimiter fields. In actuality, the differences in frame length references can be traced to the standardization process. When Ethernet was standardized by Digital Equipment Corporation, Intel, and Xerox (referred to as the DIX standard), the Version 2 specification defined a minimum frame length of 1526 bytes. That specification included all headers and trailer fields. When the IEEE subsequently standardized Ethernet as the IEEE 802.3 standard, the media access control (MAC) sublayer was only concerned with fields from the destination address through the cyclic redundancy check (CRC) carried in the four-byte frame check sequence field. Thus, the IEEE 802.3 standard recommends that the length of those fields together must range from a minimum of 64 to a maximum of 1518 bytes, which is equivalent to a minimum frame length of 72 bytes and a maximum frame length of 1526 bytes when you consider the eight bytes in the preamble and start of frame delimiter fields.
In this book we will base our Ethernet network utilization computations using minimum and maximum frame lengths of 72 and 1526 bytes, respectively. The reason for this usage is the fact that the preamble and start of frame delimiter fields always flow on a network wire and must be considered when examining Ethernet network performance.
2.1.4 Destination Address Field
The destination address identifies the recipient of the frame. Although this may appear to be a simple field, in actuality this field can vary between IEEE 802.3 and Ethernet frames with respect to field length. In addition, each field can consist of two or more subfields whose settings govern such network operations as the type of addressing used on the LAN and whether or not the frame is addressed to a specific station or to more than one station. To obtain an appreciation for the use of this field, let us examine how this field is used under the IEEE 802.3 standard as one of the two field formats applicable to Ethernet.
Figure 2.2 illustrates the composition of the source and destination address fields. As indicated, the two-byte source and destination address fields are only applicable to IEEE 802.3 networks, while the six-byte source and destination address fields are applicable to both Ethernet and IEEE 802.3 networks.
Figure 2.2: Source and Destination Address Field Formats
Although you can select either a two- or six-byte destination address field, when working with IEEE 802.3 equipment, all stations on the LAN must use the same addressing structure. Today, almost all 802.3 networks use six-byte addressing because the inclusion of a two-byte field option was primarily designed to accommodate early LANs that use 16-bit address fields.
2.1.4.1 I/G Subfield
The one-bit I/G subfield is set to 0 to indicate that the frame is destined to an individual station, while a setting of 1 indicates that the frame is addressed to more than one station. Here, the latter situation indicates a group address.
One special example of a group address is the assignment of all 1's to the address field. Here, the address, hex FFFFFFFFFFFF, which represents 48 bits all set to 1's, is recognized as a broadcast address and each station on the network will receive and accept frames with that destination address.
When a destination address specifies a single station, the address is referred to as a unicast address. A group address that defines multiple stations is known as a multicast address, while a group address that specifies all stations on the network is, as previously mentioned, referred to as a broadcast address.
2.1.4.2 U/L Subfield
The U/L subfield is only applicable to the six-byte destination address field. The setting of this field's bit position indicates whether the destination address is an address that was assigned by the IEEE ( universally administered) or assigned by the organization via software (locally administered).
2.1.4.3 Universal versus Locally Administered Addressing
Each Ethernet network interface card (NIC) contains a unique address burned into its read-only memory (ROM) at the time of manufacture. To ensure this universally administered address is not duplicated , the IEEE assigns blocks of addresses to each manufacturer. Those addresses normally include a three-byte prefix that identifies the manufacturer and is assigned by the IEEE, as well as a three-byte suffix that is assigned by the adapter manufacturer to its NIC. For example, the prefix hex 02608C identifies an NIC manufactured by 3Com, while a prefix of hex 08002 identifies an NIC manufactured by Digital Equipment Corporation, the latter acquired by Compaq Computer, which in turn was acquired by Hewlett-Packard.
Although the use of universally administered addressing eliminates the potential for duplicate network addresses, it does not provide the flexibility obtainable from locally administered addressing. For example, under locally administered addressing, you can configure mainframe software to work with a predefined group of addresses via a gateway PC. Then, as you add new stations to your LAN, you simply use your installation program to assign a locally administered address to the NIC instead of using its universally administered address. As long as your mainframe computer has a pool of locally administered addresses which includes your recent assignment, you do not have to modify your mainframe communications software configuration. Because the modification of mainframe communications software typically requires a recompile and reload activity to be performed, doing so requires the attached network to become inoperative for a short period of time. As a large mainframe may service hundreds to thousands of users, such changes are normally performed late in the evening or on the weekend , making the changes for the use of locally administered addressing more responsive to users than the changes required when universally administered addressing is used.
2.1.5 Source Address Field
The source address field identifies the station that transmitted the frame. Similar to the destination address field, the source address can be either two or six bytes in length.
The two-byte source address is only supported under the IEEE 802.3 standard and requires the use of a two-byte destination address, with all stations on the network required to be set to two-byte addressing field use. The six-byte source address field is supported by both Ethernet and the IEEE 802.3 standard. When a six-byte address is used, the first three bytes represent the address assigned by the IEEE to the manufacturer for incorporation into each NIC's ROM. The vendor then normally assigns the last three bytes to each of its NICs.
2.1.6 Type Field
The two-byte type field is only applicable to the Ethernet frame. This field identifies the higher-level protocol contained in the data field. Thus, this field informs the receiving device how to interpret the data field.
Under Ethernet, multiple protocols can exist on the LAN at the same time and Xerox served as the custodian of Ethernet address ranges licensed to NIC manufacturers as well as defining the protocols supported by the assignment of type field values. Under the IEEE 802.3 standard, the type field was replaced by a length field that precludes compatibility between pure Ethernet and 802.3 frames.
2.1.7 Length Field
The two-byte length field is only applicable to the IEEE 802.3 standard and defines the number of bytes contained in the data field. Under both Ethernet and IEEE 802.3 standards, the minimum size frame must be 64 bytes in length from preamble through FCS (frame check sequence) fields. This minimum size frame was required to ensure that there was sufficient transmission time to enable Ethernet NICs to accurately detect collisions based on the maximum Ethernet cable length specified for a network and the time required for a frame to propagate the length of the cable. Based on the minimum frame length of 64 bytes and the possibility of using two-byte addressing fields, this means that each data field must be a minimum of 46 bytes in length.
When Ethernet was developed, an interesting problem that had to be considered was the effect of collisions on short frames. For example, if a station transmitted a short frame, it was possible that transmission could be completed prior to the first bit in the frame reaching its destination. If the destination station listened to the network prior to the arrival of the first bit and assumed it was OK to transmit, a collision would result. While collisions are a common occurrence on Ethernet networks, under the preceding short frame scenario the transmitting station would incorrectly conclude that the frame was correctly received. This conclusion would occur because upon sensing an increase in voltage from the collision, the receiver, because it is closest to the collision, generates a jam signal to warn all other stations on the LAN not to transmit. However, by the time the jam signal is received by the original transmitting station, it would have falsely concluded that the previous transmitted frame was correctly received.
Due to the preceding, all frames transmitted on an Ethernet network must have a minimum length that exceeds twice the propagation delay encountered on the medium. For the original 10 Mbps coaxial cable-based LAN that has a maximum length of 2500 meters , the minimum time defined by the IEEE is 51.2 microseconds ( ¼ s). That time corresponds to 64 bytes, because 64 bytes * 8 bits/byte * (1.0 * 10 ˆ’ 7 ) seconds/bit is 51.2 ¼ s.
As the network speed increases, either the minimum frame length must increase or the maximum cable length must decrease. To provide scalability between different versions of Ethernet, frame size constraints were maintained when Fast Ethernet was developed. However, the maximum cable distance between any two stations was reduced to approximately 250 meters. When we discuss Gigabit Ethernet later in this section, we will also discuss two techniques used to make 1 Gbps operations more efficient so that a reasonable cabling distance can be supported.
2.1.8 Data Field
As previously discussed, the data field must be a minimum of 46 bytes in length to ensure that the frame is at least 64 bytes in length. This means that the transmission of one byte of information must be carried within a 46-byte data field and results in the padding of the remainder of the field if the information to be placed in the field is less than 46 bytes. Although some publications subdivide the data field to include a PAD subfield, the latter actually represents optional fill characters that are added to the information in the data field to ensure a length of 46 bytes. The maximum length of the data field is 1500 bytes, which results in the use of multiple frames to transport full screen images and almost all files transfers.
2.1.9 Frame Check Sequence Field
The frame check sequence (FCS) field is applicable to both Ethernet and the IEEE 802.3 standard and provides a mechanism for error detection. Each transmitter computes a cyclic redundancy check (CRC) that covers both address fields, the type/length field, and the data field. The transmitter then places the computed CRC in the four-byte FCS field.
The CRC is developed by treating the composition of the previously mentioned fields as one long binary number. The n bits to be covered by the CRC are considered to represent the coefficients of a polynomial M ( X ) of degree n ˆ’ 1. Here, the first bit in the destination address field corresponds to the X n ˆ’ 1 term , while the last bit in the data field corresponds to the X term. Next , M ( X ) is multiplied by X 32 and the result of that multiplication process is divided by the following polynomial:
Readers should note that the term X n represents the setting of a bit to a 1 in position n . Thus, part of the generating polynomial X 5 + X 4 + X 2 + X 1 represents the binary value 11011.
The result of the division produces a quotient and remainder. The quotient is discarded and the remainder becomes the CRC value placed in the four-byte FCS field. This 32-bit CRC reduces the probability of an undetected error to 1 bit in every 4.3 billion, or approximately 1 bit in 2 32 ˆ’ 1 bits.
Once a frame reaches its destination, the receiver uses the same polynomial to perform the same operation on the received data. If the CRC computed by the receiver matches the CRC in the FCS field, the frame is accepted. Otherwise, the receiver discards the received frame as it is considered to have one or more bits in error. The receiver will also consider a received frame to be invalid and discard it under two additional conditions. Those conditions occur when the frame does not contain an integral number of bytes and when the length of the data field does not match the value contained in the length field. Concerning the latter condition, obviously it is only applicable to the 802.3 standard because an Ethernet frame uses a type field instead of a length field.
2.1.9.1 Fast Ethernet
As previously discussed in this section, the frame format of Fast Ethernet duplicates the IEEE 802.3 frame with the exception of the use of prefix and suffix bytes that surround the frame. The prefix bit is known as the Start of Stream Delimiter (SSD), while the suffix byte is known as the End of Stream Delimiter (ESD).
The SSD is used to align a received frame for subsequent decoding while the ESD is used as an indicator that data transmission terminated normally and a properly formed stream was transmitted. Figure 2.3 illustrates how the SSD and ESD bytes are used to "frame" the IEEE 802.3 frame. At the 100 Mbps operating rate of 100BASE-TX, the frames are known as streams, which accounts for the names of the two delimiters.
Figure 2.3: Fast Ethernet Frame
In comparing Fast Ethernet to Ethernet and IEEE 802.3 frame formats previously illustrated in Figure 2.1, you will note that other than the starting and ending stream delimiters, the Fast Ethernet frame duplicates the older frames. Another difference between the two is not shown, as it is not actually observable from a comparison of frames because this difference is associated with the time between frames. Ethernet and IEEE 802.3 frames are Manchester encoded and have an interframe gap of 9.6 ¼ s between frames. In comparison, the Fast Ethernet 100BASE-TX frame is transmitted using 4B5B encoding, and idle codes are used to mark a 0.96- ¼ s interframe gap. Both the SSD and ESD fields can be considered to fall within the interframe gap of Fast Ethernet frames. Thus, computations between Ethernet/IEEE 802.3 and Fast Ethernet becomes simplified as the latter has an operating rate ten times the former and an interframe gap one tenth the former.
2.1.9.2 Gigabit Ethernet
As previously discussed in this chapter, an increase in the speed of an Ethernet network must be accompanied by either an increase in the minimum frame length or a decrease in the maximum cable length. At a data rate of 1 Gbps, maintaining a minimum length frame of 64 bytes (72 when the preamble and start of frame delimiter fields are considered) would reduce the maximum network diameter to approximately 20 meters. While this distance might be suitable for connecting a hub to a group of stations within close proximity of one another, it is not suitable to support horizontal wiring within a building where a 10-meter distance is allowed from a wall faceplate to the desktop. To enable Gigabit Ethernet to support a network diameter of up to 200 meters, a technique referred to as Carrier Extension was added to the technology.
2.1.9.3 Carrier Extension
Carrier Extension results in an extension of the Ethernet slot time from 64 bytes (512 bits) to a new value of 512 bytes (4096 bits). To accomplish this extension, frames less than 512 bytes in length are padded with special carrier extension symbols. Note that under Gigabit Ethernet, the minimum frame length of 64 bytes is not changed. All frames less than 64 bytes in length are first padded out to a minimum of 64 bytes. The carrier signal placed on the network is then extended to provide a minimum carrier length of 512 bytes.
The preceding discussion of frame length is based on the IEEE use of terminology and does not consider the eight bytes associated with the preamble and start of frame delimiter fields. Figure 2.4 illustrates the Gigabit Ethernet frame format, to include the location where non-data symbols are added. Note that the FCS is calculated only on the original, nonextended frame. At the receiver, the extension symbols are removed before the FCS value is checked. When we examine the performance of Gigabit Ethernet, we add eight bytes to the preceding values to account for the preamble and start of frame delimiter because both flow on a wire.
Figure 2.4: Gigabit Ethernet Frame Format with Carrier Extension
Carrier Extension is only applicable to half-duplex transmission. This is because full-duplex transmission uses the collision detection wire pair for transmission in the opposite direction since collisions cannot occur on a full-duplex connection. As we note later in this chapter, Carrier Extension can significantly degrade performance associated with short-packet transmission. In an attempt to offset short-packet performance, a second technique, referred to as Packet Bursting, is employed by Gigabit Ethernet.
2.1.9.4 Packet Bursting
Packet Bursting represents a Gigabit Ethernet technique developed in an attempt to compensate for performance degradation associated with Carrier Extension.
Under Packet Bursting, a station with more than one frame to transmit can transmit multiple frames if the first frame is successfully transmitted. If the first frame is less than 512 bytes in length, Carrier Extension is applied to that frame. Succeeding frames in the burst are limited to the length of the maximum frame length. That is, a burst can vary between a maximum of thirteen 64-byte frames to one 1518-byte frame.
Figure 2.5 illustrates the effect of Gigabit Packet Bursting. In this example, note that the first two packets transmitted were less than 512 bytes in length and were extended. Because Packet Bursting is controlled by a burst timer that expires after a duration of 1500 bytes, any packet being transmitted during the burst time will continue to completion. This is indicated by packet 3. Also note that an interframe or interpacket gap occurs between packets when Packet Bursting occurs. However, because transmission occurs at 1 Gbps, the gap between frames is reduced from 9.6 ¼ s on a 10 Mbps Ethernet network to 0.096 ¼ s.
Figure 2.5: Gigabit Ethernet Packet Bursting
2.1.10 Frame Overhead
As previously indicated in this section, each Ethernet frame consists of six fields (if we consider the preamble and start of frame delimiter as one field and do not consider the SSD and ESD fields, as they are considered to fall within the interframe gap), of which only one field actually transports information. That data field must contain a minimum of 46 bytes even if the frame is transporting a single character response to a client/server query. Thus, the overhead associated with an Ethernet frame depends on both the fixed length of the frame fields that do not carry information as well as the number of characters carried in the data field, which can vary from one to 1500 bytes.
If we assume six-byte addressing, which is applicable to almost all modern Ethernet networks, the number of fixed overhead bytes per frame is 26, consisting of eight preamble bytes, six destination and six source address bytes, two bytes for the type or length field, and four bytes for the FCS field.
A one-byte response carried in the data field must be padded by the addition of 45 fill characters when six-byte addressing is used. In this situation, the overhead required to carry a one-byte character is 26 + 45, or 71 bytes.
Now consider the situation in which you have 46 bytes of data to transmit. Here, the 46 bytes would not require the addition of pad characters because the frame length when using six-byte addressing would be 64 bytes (72 when considering the preamble and start of frame delimiter fields), which is the minimum frame length. Thus, 64 bytes of data would result in a frame overhead of 26 bytes.
Table 2.1 summarizes the overhead associated with transporting information in Ethernet frames as the number of bytes of information varies from one to the maximum 1500 bytes the frame can carry. As indicated in Table 2.1, the overhead associated with transporting information within an Ethernet frame can vary considerably, ranging from a high of 98.61 percent to a low of 1.7 percent when the maximum length data field is used to transport information.
| Information Carried in Data Field (bytes) | Ratio of Bytes of Frame Overhead to Frame Length | Percent Overhead |
|---|---|---|
| 1 | 71/72 | 98.61 |
| 10 | 62/72 | 86.11 |
| 20 | 52/72 | 72.22 |
| 30 | 42/72 | 58.33 |
| 45 | 27/72 | 37.50 |
| 46 | 26/72 | 36.11 |
| 64 | 26/90 | 28.89 |
| 128 | 26/154 | 16.88 |
| 256 | 26/282 | 9.22 |
| 512 | 26/538 | 4.83 |
| 1024 | 26/1050 | 2.48 |
| 1500 | 26/1526 | 1.70 |
2.1.10.1 Gigabit Ethernet Overhead
The previous computations did not take into consideration the overhead associated with frames transmitted on a Gigabit Ethernet network. As previously noted in this chapter, packets less than 512 bytes in length (not including their eight-byte header) are extended in length through the use of carrier extension symbols. This means that a packet transporting a one-byte data character first is extended by the addition of 45 padding bytes and then further extended 448 carrier extension symbols. Thus, a Gigabit Ethernet frame transporting one data character contains 520 bytes, to include preamble and start of frame delimiter fields, of which 519 represent overhead.
To illustrate some additional overhead computations, let us assume a Gigabit Ethernet frame is transporting 64 bytes of data. In this case, there is no requirement for padding characters. However, because the frame must be 512 characters in length without considering the eight-byte header, this means there must be 430 carrier extension symbols appended to the frame (448 ˆ’ 18). If you add the normal 26 overhead bytes to the 430 carrier extensions, you obtain a total overhead of 456 bytes.
Table 2.2 summarizes the overhead associated with transporting information in Gigabit Ethernet frames as the number of bytes of information varies from one to the maximum of 1500 that the frame can transport. The overhead entries in Table 2.2 indicate that the performance of Gigabit Ethernet can be significantly degraded by interactive transmission. Although Packet Bursting is designed to counter the effect of carrier extensions, in reality, interactive query-response applications will not benefit from Packet Bursting. This is because a short query is a one-time event, followed by a response and then perhaps another query, activity that Packet Bursting cannot enhance.
| Information Carried in Data Field (Bytes) | Ratio in Bytes Frame Overhead to Frame Length | Percent Overhead |
|---|---|---|
| 1 | 519/520 | 99.81 |
| 10 | 510/520 | 98.07 |
| 20 | 500/520 | 96.15 |
| 30 | 490/520 | 94.23 |
| 40 | 480/520 | 92.31 |
| 45 | 475/520 | 91.35 |
| 46 | 474/520 | 91.12 |
| 64 | 456/520 | 87.69 |
| 128 | 392/520 | 75.38 |
| 256 | 264/520 | 51.56 |
| 384 | 136/520 | 26.15 |
| 512 | 26/538 | 4.83 |
| 1024 | 26/1050 | 2.48 |
| 1500 | 26/1526 | 1.70 |
One question you might have by now is: How can you use the overhead information presented in Tables 2.1 and 2.2 to your advantage for increasing network performance? The data contained in those tables can be extremely important for network performance if your organization is developing client/server applications. For example, assume your programming staff is developing screens to be displayed on a user 's workstation in which the client enters information that is then transmitted to the server to initiate different activities. If there is a choice between generating a screen display that results in the transmission of a small number of characters to the server or generating a lesser number of screen displays and transmitting the fields of several screens at one time, the latter is preferable as it results in a lower overhead. This, in turn, will reduce the number of frames carried on the network. Because each version of Ethernet requires a fixed time gap between frames, reducing the number of frames increases the efficiency of dataflow. While altering the transmission of data carried by frames will probably not result in any noticeable increase in network performance if your network only has a handful of workstations, more notable results can be expected to be observed if your network has a large number of stations that perform client/server operations on a regular basis.
EAN: 2147483647
Pages: 111