 | My Personal Bookstore & Wishlist |
|
IEEE LAN’s
Access method’s (polling, token passing of contention): This method decides the presentation and possibilities from the network • Polling: Making periodic requests is called polling. Polling also reduces the burden on the network because the polls originate from a single system are at a predictable rate. The shortcoming of polling is that it does not allow for real-time updates. If a problem occurs on a managed device, the manager does not find out until the agent polled. Mostly used in a star network topology. • Token passing: Token passing that every device on the network receives a periodic opportunity to transmit. The token consists of a special frame that circulates from device to device around the ring. Only the device that possesses the token is permitted to transmit. After transmitting, the device restarts the token, enabling other devices the opportunity to transmit. • Contention (CSMA/CA of CSMA/CD): A condition occuring in some LAN’s wherin the Media Access Control sublayer allows more than one node to transmit at the same time, risking collisions. Mostly used in a bus network topology. Architecture of the IEEE 802 Standards: Network type IEEE 802.2: Defines the LLC sublayer protocol. Network type IEEE 802.3: Network with a bus-topology and the access method CSMA/ CD, 10 Mbps. Defines the MAC and physical layer for CSMA/CD. Network type IEEE 802.4: Network with a bus-topology and the access method token passing, 2.5 Mbps. Network type IEEE 802.5: Network with a ring-topology and the access method token passing, 4 Mbps. Defines the MAC and physical layer for a Token Ring network. This sublayer provides a network interface to Upper-Layer Protocols (ULP) and is concerned with transmitting data between two stations on the same network segment. An interface between the LLC sublayer and upper-layer protocols is a Link Service Access Point (LSAP). It is a logical address that identifies the upper-layer protocol from which the data originated or to which the data should be delivered. LLC Delivery Service: Was designed to provide a variety of delivery services, which determine the level of communication integrity established between devices. LCC support the following three types of delivery service: • Type 1 service, Unacknowledged Datagram Service (UDS), supports point-to-point, multipoint, and broadcast transmission. Does not perform error detection and recovery or flow control. • Type 2 service, Virtual Circuit Service (VCS), provides frame sequencing, flow control, and error detection and recovery. • Type 3 service, Acknowledged Datagram Service (ADS), implements point-to-point datagram service with message acknowledgements, and functions somewhere between type 1 and type 2 service. Devices have a limited number of receive buffers, used to store frames that have been received but not processed. If the sending device continues to transmit while the destination receive buffers are full, frames not received are lost. Flow control ensures that frames are not sent at a rate faster than the receiving device can accept them.
Figure 39 shows the receiving computer risks losing data whenever its communication buffers become full. A variety of mechanisms can be used to provide flow control: The simple stop-and-wait method requires the receiver to acknowledge received frames, signalling a readiness to accept more data. This mechanism is suitable to a connectionless, datagram service. If the sender must wait for an acknowledgement of each frame, multiframe transmissions are handled inefficiently. The more sophisticated sliding-window technique enables the sender to transmit multiple frames without waiting for an acknowledgement. The receiver can acknowledge several datagrams at one time. A window determines the number of frames that can be outstanding at a given time, ensuring that the receiver's buffer do not overflow. The complexity of sliding-windows flow control requires a connection-oriented LLC service. Error detection is performed at the MAC layer, but error recovery, when performed at the data link layer, is a function of LLC.
Data-communication processes allocate memory, commonly known as communication buffers, for the sake of transmission and reception of data. Communication buffers serve as holding areas where inbound data traffic is temporarily kept for subsequent handling by the CPU. Depending on the rate at which incoming data is handled by other components of the communication process, the communications buffers often become full. A computer whose communications buffers become full while still in the process of receiving data runs the risks of discarding extra transmissions and losing data unless a data flow control mechanism is employed. A proper data flow control technique calls on the receiving process to send a stop sending signal to the sending computer whenever it cannot cope with the rate at which data is being transmitted. The receiving process later sends a resume sending signal when data communications buffers become available. LLC Data Format: The LLC layer constructs a PDU by appending LLC-specific fields to the data received from upper layers.
Figure 40 shows the format of the LLC protocol data unit. The fields in figure 40 are as follows: • The Destination Service Access Point (DSAP) address that identifies the required protocol stack on the destination computer. • The Source Service Access Point (SSAP) address associated with the protocol stack that originated the data on the source computer. • The Control Information that varies with the function of the PDU. • The Data received from upper-layer protocols in the form of the network layer PDU. • Medium Access Control (MAC): This sublayer provides the method by which devices access the shared network transmission medium. Physical device addresses are defined at the MAC protocol sublevel. Physical addresses, therefore, frequently are referred to as MAC addresses.
Figure 41 shows the format of an IEEE 802 MAC address. The bit’s 46 and 47 in figure 41 are as follows: • Bit 47 is the Physical/Multicast bit. If the bit is 0, the address specifies the physical address of one device on the network. If the bit is 1, it specifies a multicast address that identifies a group of devices. • Bit 46 is the U/L bit and indicates whether the address is universally or locally administrated. If the bit is 0, universally administrated address. If the bit is 1, locally administrated address.
Figure 42 shows IEEE 802 standards related to the OSI reference model. Utilise the same CSMA/CD access control mechanism that was developed for Ethernet II. The same media-signalling techniques are employed and 802.3 and Ethernet II network hardware are interchangeable. 802.3 and Ethernet II frames may be multiplexed on the same media. The primary difference between the 802.3 and Ethernet II standards has to do with frame formats.
Figure 43 shows the schematic of an Ethernet network. Typically, local area networks permit a single node to transmit at a given time. Access control methods are systems that enable many nodes to have access to a shared network medium by granting access to the medium in an organised manner. Ethernet uses an elegant access control method, called carrier sence. When a node has data to transmit, it senses the medium, essentially listening to see if any other node is transmitting. If the medium is busy, the node waits a few microseconds and tries again. If the medium is quiet, the node begins to transmit. The full name for this approach is Carrier Sence Multiple Access (CSMA), permitting multiple nodes to access the medium through a carrier sence method. Carrier Sence Multiple Access/Collision Avoid (CSMA/CA): The listen to the wire to check if there is someone that wants to communicates, the pronounce that the are ready to start with a communication (burst). When two termi-nals on the same moment are ready to start with a communication then the commu-nication will be delayed for a random time by both terminals. Carrier Sence Multiple Access/Collision Detection (CSMA/CD): The start with there communication when the think that the are the only ones that wants to communicate. When after a searten time seams that the don't where the only ones that wants to communicate, both terminals stops there communication for a random time before the trey again. With a much better rendement then a token that needs to pass all the different terminals offers the CSMA/CD method the disad-vantage that it is not possible to now exactly which response time they need to use with a danger for saturation if there is much intensive traffic. Before the stations can send the need to do next 5 steps on a CSMA/CD-network: 1 - listen to the wire before the can send, 2 - wait if the cable isn't free, 3 - send and listen to the wire to check if there are collisions, 4 - if there is a collisions, wait again before you can send it again, 5 - send it again or cancel it. Before the stations can recieve the need to do next 4 steps on a CSMA/CD-network: 1 - inspectation of the incoming packets and checking on fragmentation, 2 - read and check the destination address, 3 - when the packet is for the local station, check the packet to sea if it's intact, 4 - process the packet. A brief period of time must expire before a transmitted electrical signal reaches the furthest extents of the medium on which it is sent. As the two signals flow through the medium, eventually they overlap in an event called a collision. Collisions always damage data, and having a mechanism for dealing with collisions when they occur is of paramount importance. Ethernet nodes detect collisions by continuing to listen as they transmit. If a collision takes place, the nodes measure a signal voltage that is twice as high as expected. After detecting a collision, the nodes transmit a jamming signal that notifies all nodes on the network that a collision has occurred and the current frame should be disregarded. Then the nodes wait random amount of time before attempting to retransmit. Because each node delays for a different time, the likelihood of a new collision is reduced. This technique of managing collisions is called Collision Detection (CD), making the complete abbreviation for the Ethernet access control method CSMA/CD. Collisions are part of the normal operation of an Ethernet. Because CSMA/CD is an exceptionally efficient access control method, normal collision activity does not seriously affect network performance. They occur when two or more systems transmit at the same time contending for the right to control the network. If a system transmit 64 bytes, it is considered to be in control, and the other systems are supposed to be quiet until the controlling system has finished. It is possible, if the total length of an Ethernet exceeds the specifications, for a system not to know that another system has control of the network and to transmit right over the controlling system's packet. This creates a packet greater than 64 bytes long with a CRC error. The busier the network, the more this problem becomes.
Figure 44 shows collisions on an Ethernet. Sometimes when an installation doesn't work because the cable is to long or otherwise out of specification, people use a transceiver or network card that functions even over an out-of-specification link to solve the problem. Don't do it. You are not solving the problem. You're just hiding the problem that may came back to haunt you in the future. In a large 10BASET installation, hubs that can be remotely managed are almost indispensable. Simple Network Management Protocol (SNMP) is the standard management software for TCP/IP networks. The agent is the software that reports information about a device back to the management station. SNMP may help you manage the PC’s on your network. Late collisions are undetected collisions caused by a cable segment that is too long and are one example of why you'll regret violating the Ethernet specifications. Ethernet II Frames:
Figure 45 shows the structure of an Ethernet II frame. • The minimum length of an Ethernet frame is 6+6+2+46+4=64 octets • The maximum length of an Ethernet frame is 6+6+2+1500+4=1518 octets The fields in figure 45 are as follows: • The preamble consists of a series of 8 bits in a specific pattern that notifies receiving nodes that a frame is beginning. The preamble begins with seven octets (8-bit groups, frequently referred to as byte) of the pattern 10101010. The final octet of the preamble has the bit pattern 10101011. The purpose of the preamble is to signal the beginning of a frame, and the preamble is not formally part of the frame. Therefore, the octets in the preamble are not counted as part of the length of the frame. • The destination and source address each consist of 48 bits (6 octets). Each node on the network is assigned a unique 48-bit address. This information enables receiving nodes to identify frames that are addressed to them, and also enables the receiver of a message to reply to the sender. • The type field (EtherType) is a 16-bit (2 octets) field that designates the data type of the data field. The EtherType enables the network drivers to demultiplex the packets and direct data to the proper protocol stack. The type mechanism enables Ethernet networks to support multiple protocol stacks. • The data field contains the Protocol Data Unit (PDU) received from upper-layer protocols. For TCP/IP its constructed of three components: The IP header, the TCP header, and the application data. The length of the data field can bee from 46 to 1500 octets, inclusive. If the data field is less than 46 octets in length, upper-layer protocols must pad the data to the minimum length. • The Frame Check Sequence (FCS) is a 32-bit code that enables the receiving node to determine if transmission errors have altered the frame. This code is derived through a Cyclic Redundancy Checksum (CRC) calculation which processes all fields except the preamble and the frame sequence. This CRC value is recalculated by the receiving node. If the CRC calculation by the receiver matches the value in the FCS, it is assumed that transmission errors didn’t occur. Ethernet II Node Address: Consist of 48 bits, organised in three fields, commonly organised in sec octets, six groups of 8 bits.
Figure 46 shows the structure of an Ethernet II Node Address. • Bit 47 is the Physical/Multicast bit. If the bit is 0, the address specifies the physical address of one device on the network. If the bit is 1, it specifies a multicast address that identifies a group of devices. Vendors are assigned unique vendor codes that are used to identify their adapters. This registration system ensures that each Ethernet device that is manufactured has a physical address that is unique in the entire world. The Globally Administrated Address is designated by the manufacturer of the Ethernet equipment. Because each manufacturer is assigned a unique vendor ID, and the manufactures assign a different identification number to each equipment produced, the complete Ethernet ID for each Ethernet device is unique. Ethernet wiring comes in three forms: • Thicknet : IEEE 10BASE5 standard, coax cable .5" diameter, used for backbone Ethernet to interconnect other networks • Thinnet : IEEE 10BASE2 standard, coax cable .2" diameter, used to directly connect PC’s • UTP : IEEE 10BASET standard, used to directly connect PC’s, these systems requires a concentrator or hub to operate. Ethernet wiring limits: Max. 10BASE5 10BASE2 10BASET Segment length 500 m 185 m 500 m Repeaters or concentrators 4 4 4 Total length 2500 m 925 m 2500 m Nodes per segment 100 30 512 Workstation cable N/A N/A 100 m Each of the cable standards has a three-part name. The first number indicates the data rate in megabits per second. BASE specifies baseband operation, and BROAD indicates a broadband network. The final designation suggest the cable type. • 10BASE5 : Thick, 50-ohm coaxial cable. • 10BASE2 : Thinner coaxial cable. • 10BASE-T : UTP cable. • 10BROAD36: A broadband cable system that enables multiple 10 Mbps channels to be carried by the same coaxial medium. • 100BASE-TX: Utilises two pairs of high-grade UTP cable, 100 Mbps. • 100BASE-T4: Utilises four pairs of standard grade UTP cable, 100 Mbps • 100BASE-TF: Utilises optical fibre, 100 Mbps.
Figure 47 shows the format of a IEEE 802.3 Frame. • The minimum length of an IEEE 802.3 frame is 6+6+2+46+4=64 octets. • The maximum length of an IEEE 802.3 frame is 6+6+2+1500+4=1548 octets. The fields in figure 47 are as follows: • The preamble consists of a series of 8 bits in a specific pattern 10101010. • The Start Frame Delimiter (SFD) is a one octet with the bit pattern 10101011. • The destination and source address each consist of 48 bits (6 octets). Each node on the network is assigned a unique 48-bit address. This information enables receiving nodes to identify frames that are addressed to them, and also enables the receiver of a message to reply to the sender. • The length field consists of 2 octets that specify the number of octets in the LLC data field. This value must be in the range 46 through 1500, inclusive. • The LLC data field contains the Protocol Data Unit (PDU) received from the LLC sublayer, consisting of the LLC header and data. The size of this field can be from 46 to 1500 octets, inclusive. If the data field is less than 46 octets in length, upper-layer protocols must pad the data to the minimum length. • The Frame Check Sequence (FCS) is a 32-bit code that enables the receiving node to determine if transmission errors have altered the frame. This code is derived through a Cyclic Redundancy Checksum (CRC) calculation which processes all fields except the preamble and the frame sequence. This CRC value is recalculated by the receiving node. If the CRC calculation by the receiver matches the value in the FCS, it is assumed that transmission errors didn’t occur. • Implementing TCP/IP over IEEE 802.3:
IEEE 802.5 Token Ring is the second most commonly employed LAN physical layer, trailing significantly behind Ethernet. Each time a device needs to transmit, some probability exists that the network will be busy. And, even when the device successfully begins to transmit, some probability exists that another device will also transmit and cause a collision, forcing both devices to back off and try again. These probabilities increase as the network becomes busier, until a point is reached at which a device needing to transmit data becomes extremely unlikely to receive the opportunity to do so. Because network access on a CSMA/CD network is uncertain, CSMA/CD is called a probabilistic access method. The mere probability of access is unacceptable in certain critical situations such as industrial control. Suppose that an overheat urgently needs to send a warning to the factory operators. If even a possibility exists that the sensor cannot access the network, the factory designers will not take the situation lightly. Token access guarantees that every device on the network receives a periodic opportunity to transmit.
Figure 49 shows the token access method in a ring network. The token consists of a special frame that circulates from device to device around the ring. Only the device that possesses the token is permitted to transmit. After transmitting, the device restarts the token, enabling other devices the opportunity to transmit. The initial 4 Mbps implementation of Token Ring permitted a single token to circulate on the network. Before releasing a token on the network that enabled other devices to transmit, a device that transmitted a frame waited for the frame to return after circulating the ring. A new feature, called Early Token Release (ETR), introduced with the newer 16 Mbps Token Ring, enables a sending device to release a token immediately after it completes transmission of a frame. Thus a token can circulate at the same time as a data frame. Although token access control appears simple, numerous problems lie beneath the surface. The point of introducing them is to illustrate that the control mechanisms Token Ring uses are significantly more complicated than those required for CSMA/CD. These control mechanisms take up network bandwidth, reducing the efficiently of Token Ring. To compensate for this added complexity, Token Ring offers significant benefits. Data throughput of a Token Ring can never reach zero, as is possible with an Ethernet experiencing excessive collisions. Although network performance slows as demand increases, every device on the network receives a periodic opportunity to transmit. Token Ring possesses a capability to set network access priorities, which is unavailable in Ethernet. High-priority devices can request preferred network access. This capability enables a critical device to gain greater access to the network. Token Ring was also designed to provide a higher level of diagnostic and management capability than is available with Ethernet. The mechanisms that compensate for Token Ring errors provide a capability for diagnosing other network problems, as well. For example, detecting devices causing network errors and forcing those devices to disconnect from the network, is possible. Also, in the cabling system IBM designed, the network is services by two rings of cable. In the event of a cable break, using the media ring to reconfigure the network and keep it operating is possible. Nevertheless, Ethernet remains the most popular network physical layer. Ethernet works well in the majority of networks and costs considerably less than Token Ring. Equipment for Token Ring costs two-to-three times as much as corresponding Ethernet components.
Figure 50 shows how Token Rings are wired in a star. • Several reasons can be cited for Token Ring's lower popularity: • It was developed as an IBM technology. Although Token Ring technology is now offered by great many vendors, many in the user community perceive it as proprietary. • Ethernet is simple, reliable, and effective for the majority of networks, and at the same time, cost significantly less than Token Ring. • TCP/IP has traditionally been wed to Ethernet II. Growing industry demand for TCP/IP has accompanied a recent surge in the Ethernet popularity. Nevertheless, Token Ring is an effective physical layer technology with features that make it preferable under some circumstances.
Figure 51 shows the format of a Token Ring frame. Three major sections can be specified, as follow: • Start-of-Frame Sequence (SFS): This section signals the network devices that a frame is beginning. • Data section: This section contains control information, upper-layer data, and that a frame is beginning. • End-of-Frame Sequence (EFS): This section indicates the end of the frame and includes several control bits. The fields in figure 51 are as follows: • The Starting Delimiter (SD) field is a single octet that consists of electrical signals that cannot appear elsewhere in the frame. The SD violates the rules for encoding data in the frame and contains nondata signals. • The Access Control (AC) field includes priority and reservation bits used to set network priorities. It also includes a monitor bit, used for network management. A token bit indicates whether the frame is a token or a data frame. • The Frame Control (FC) field indicates whether the frame contains LLC data or is a MAC control frame. Several types of MAC frame are used to control network functions. • The Destination Address (DA) specifies the station or stations to which the frame is directed. Multicasts and broadcasts are possible in addition to transmission to a single device. 16- and 48-bit addresses are supported. • The Source Address (SA) specifies the device that originated the frame. The DA and SA address must utilise the same format. • The Information field contains LLC data or control information if it appears in a MAC control frame. • The Frame Check Sequence (FCS) is a 32-bit cyclic redundancy check that is applied to the FC, DA, SA, and information field. • The Ending Delimiter (ED) violates the network data format and signals the end of the frame. This field includes two control bits. The intermediate bit indicates whether this is an intermediate or the final frame in a transmission. The error bit is set by any device that detects an error, such as in the FCS. • The Frame Status (FS) field contains other control bits that indicate that a station has recognised its address and that a frame has been copied by a receiving device.
|
