Review of Last Class 1. Numbering Systems 2. Code Sets 3. Serial/Parallel 4. Async/Sync 5. Protocols and encapsulation 6. Layering and OSI Reference Model a. Physical b. Data link c. Network d. Transport e. Session f. Presentation g. Application 7. Standards organizations 8. PSTN example Review Homework Administrative Start thinking about project topic For attendance/participation credit: Computer lab 6th floor Sign up required Lightstream 1010 (ATM), Cisco Cat 5000, Cisco router (model ?), Linux, Windows NT (PDC/BDC), NetWare 3.12/5.0 DePaul IEEE society http://www.depaul.edu/~ieee/ DePaul Networks Research Seminar http://bach.cs.depaul.edu/brewster/seminar/ Weblinks (skim only) http://www.andrewboon.com.au/html/BOONSCS1.html comp.dcom.cabling Structured Cabling guides: http://www.anixter.com/techlib/standard/cabling/ General info: http://www.wiring.com http://www.cabletesting.com/ Telecommunications presentation: http://www.moultonco.com/semnotes/telecomm/teladd.htm Read: http://www.networkuptime.com/faqs/token-ring/index.html II. Layer One Technology (Physical) A. Signal and Spectrum 1. Analog and Digital Encoding An analog signal is continuous in time and value. A digital signal is discrete in time and value. Modulation a. Frequency b. Amplitude c. Phase 2. Undesirable Effects a. Noise - external or internal b. Attenuation - signal weakens with distance c. Interference - external or internal (crosstalk) d. Distortion - imperfections in the channel 3. Fundamental Limitations a. Physical i. bandwidth ii. noise/power b. Technological i. hardware availability ii. regulatory/health B. Transmission Media Characteristics 1. Copper Cabling a. Twisted Pair i. Twists ii. Sheilding iii. Category 1-5 Connectors: 8-pin modular jack (RJ-45), RJ-11 Solid core versus stranded. Connector differences. Sample cables, connectors and test equipment b. Coaxial 2. Optical Fiber a. Multi-mode b. Single-mode Connectors: SC, ST, MIC, RJ-MT Sample cables, connectors, shine light 3. Wireless 4. Media Testing The installed links need to be tested to ensure transmission performance characteristics can be meet for current and future needs. Note: we are primarily dealing with copper and fiber cabling systems in this class. Horizontal cable Vertical risers Patch panels Patch cords a. ANSI/EIS/TIA 568-A Commercial Building Wiring Standard Generic telecommunication wiring system which will support multi-product, multi-vendor environments. Category 5 (100MHz) Category 4 (20MHz) - rare Category 3 (16MHz) - still common Note: megahertz is not equal to megabits/second. It's the difference between signalling and encoding. ex: 10BASE-T requires 10MHz, 2 pairs, runs at 10Mb/s 100BASE-T requires 80MHz, 2 pairs, runs at 100Mb/s b. TSB-67 Defines performance specifications for installed UTP cabling links. - Basic Link - Channel Microtest PentaScanner demo Straight-thru and cross-over cables C. Repeaters Hardware device which performs signal re-generation from one physical link (medium) to another of the same type. Physical layer device. Does not understand frames. 1. Amplify signal 2. Re-generate Note: Repeaters often also amplify imperfections (noise) along with the data signal. There are physical limitations in using repeaters, notably distance. D. Hubs A multi-port repeater which connects multiple physical links (attached computers or other hubs) to a shared transmission medium ex: Ethernet Note: vendors often build upper layer intelligent functions into these devices (hubs/repeaters) to extend their functionality. ex: MAC address filtering, IP address for net management 10BASE-T hub demo E. Physical Network Topologies 1. Point to point 2. Bus 3. Ring 4. Star 5. Mesh 6. Tree 7. Hybrid III. Packets versus Circuits Note: In computer networks we typically think about packets (IP), but often these packets are transmitted over physical circuit based networks. A. Circuit Switching A dedicated path between two communicating nodes. Often a physical connection made up of TDM slots or FDM bands. TDM: Time-division multiplexing Simultaneous transmission of multiple signals over the same channel, separated by the time domain. ex: [channel 24]...[channel 2][channel 1] start ---> data flow time n+24 ... time n+2 time n+1 time 0 FDM: Frequency-division multiplexing Simultaneous transmission of multiple signals over the same channel, separated by the frequency domain. ex: X Hz ---> [channel 1] ---> data flow Y Hz ---> [channel 2] ---> data flow Z HZ ---> [channel 3] ---> data flow etc. time n 1. Guaranteed bandwidth 2. Connection setup/teardown Note: Unused channel bandwidth is wasted, it cannot be re-allocated to other channels/users. Performance (latency, loss, order) is constant. No surprises. If network is overloaded, connections are refused. B. Packets Transmitting station sends packet/message to the network addressed to the destination. The network forwards the packet/message along intermediate nodes (packet switches) until it reaches it's destination. 1. No dedicated path, packets/messages may take several alternative routes. 2. Performance (latency, loss, order) may change due to network conditions (load). If network is overloaded, messages are either slowed or dropped (randomness). Note: Multiple stations sending packets to the network can be viewed as a chaotic form of time-division multiplexing. It lacks the reserved time slots. How do we provide fairness? Stay tuned! Why do we use packet/circuit switching at all? Because we cannot directly connect to everything! Packet/Circuit switches typically have multiple input/output ports to computers and other networks. Sample network diagram demonstrating switching. IV. Layer Two Technology (Data Link) Remember that this layer is responsible for such details as packet framing, addressing (hardware), medium access, error control, and flow control over the physical layer. Note: This layer is often associated with LANs, but all network technologies (i.e. WANs) would require and define at least some of these functions as well. We will primarily explore this layer as it applies to LAN technology, covering WAN architectures later in the course. A. Medium Access Control (MAC) Note: We are primarily concerned with multistation access to a shared medium/network for this discussion. 1. Centralized access scheme A master node coordinates access to the network and between communicating nodes. a. Greater control for security, priorities, guaranteed capacity b. Relatively simple access logic at at each station c. Single point of failure Note: This can be overcome by an election process, but this imposes complexity d. Delay imposed by requiring master's involvement in transmission ex: cellular phone network techniques: circuit mode, packet (polling/reserveration) 2. Distributed access scheme All nodes are peers and can communicate directly, but must collectively participate in a access control function. a. Lower message delays b. Distributed nature provides for reliability c. Allow for high network utilization (cost effective) d. Distributed coordination can be problematic i. unfair nodes ii. inter-operability issues e. Complexity is increased ex: Ethernet techniques: round robin (token passing), contention (CSMA) MAC FRAME FORMAT The encapsulated data closest to the physical layer contains MAC frame information. We will discuss specific frame formats when we discuss Ethernet and Token Ring technology. MAC control fields: priority, reservation Destination MAC address: the physical attachment point on the data link network (usually LAN) where this frame should be delivered to Source MAC address: the physical attachment point of the sending station on the data link network MAC addresses are typically 48 bits (6 bytes) in length. Burned in addresses (UAA) contain two parts, the OUI (3 bytes) and a unique code from the manufacturer in the last 3 bytes. LAA's can also be used. Reservered addresses are used for specific purposes (i.e. broadcast and funtional addresses). LLC/Type: Information which specifies the next protocol layer (either LLC or the network layer information) LLC: Logical link control provides error control (using ACKs) if necessary and flow control if necessary. Often this sub-layer of the data link layer is not used. In addition the LLC if used, will specifiy the upper layer protocol via service access points (a form of multiplexing) and connection/connection-less oriented communications. Note: Because LAN technologies are inherently very reliable and in practice LLC is only used in specific environments (i.e. SNA) we will spend our time with these similar types of services that are often performed in the tranport layer (i.e. TCP). CRC: Cyclic redundancy check is used for error detection only. A packet of data is treated as a long binary number. The number is divided by a unique prime number and the remainder serves as the "redundant digits". The size of the prime number determines the size of the redundant digits. The CRC is attached to a data packet and transmitted. The receiver compares the remainder to it's own calculation to determine if a error has occurred. B. Logical Link Control (see above - most of the interesting stuff is in the MAC sub-layer) For more on LLC see Chapter 5 of Local & Metropolitan Area Networks, William Stallings, ISBN: 0-13-190737-9 C. Traditonal LANs Specific implementations of physical/MAC layer technologies. 1. Token Ring/802.5 A layer 1/2 star wired, logical ring based technology which employs a shared medium arbitration method known as token passing. IBM choose the token ring as it's LAN technology of choice in the early 1980's. At the time it was a more robust and reliable LAN transmission technique compared to 10BASE-2/10BASE-5 Ethernets. However, when 10BASE-T came out, it mostly spelled the eventual death for token ring networks. Ethernet was cheaper, easier and more widely deployed. We will discuss these advantages and disadvantages of token ring verus ethernet in the next class. a. The token frame i. 3 byte frame which circles the ring ii. stations that need to transmit must first capture the token b. The active monitor (AM) (aka ring monitor) i. Designated station on the ring which actively monitors the ring ii. Responsible for ensuring token is available and initiates neighbor notification c. Station 5-phase insertion process 0. Lobe test (adapter to hub/MAU/switch). Station sends 2000 1500 byte test frames to test the lobe (for a break) and the receive logic of the adapter. 1. Monitor check. A timer is started at the station while it watches for 1 of 3 types of frames: active monitor, standby monitor or ring purge. If none of these frames are seen, the station will initiate the token-claiming process. 2. Duplicate address (MAC) check. The station sends out at least two frames with it's own MAC address in the source and destination fields of the token ring frame. When the frame travels around the ring, back to the sender, it checks the address recognized bit and frame copied bit. If either of those bits are set, it knows another station on the ring is using it's MAC address. 3. Participate in Neighbor Notification. Station will receive it's Nearest Active Upstream Neighbor (NAUN) address and also identify itself to the next station downstream. If a station tries to insert into a beaconing ring, this is the phase at which it will fail. 4. Request initialization. The station requests any changed parameters (from the default) for the ring by the ring parameter server (RPS) via a reserved functional address. This device is usually a bridge or switch which may contain ring number and soft error timer parameters. d. 802.5 frame format 1. Starting delimiter (1) 2. Access control (1) i. priority (3) ii. token (1) iii. monitor (1) iv. reservation (3) 3. Frame control (1) i. frame type (2) ii. control (6) Determine LLC layer info or MAC protocol control 4. Destination address (2 or 6) 5. Source address (2 or 6) 6. Data (variable) 7. Frame check sequence - CRC (4) 8. Ending delimiter (1) 9. Frame status (1) i. address recognized (1) ii. frame copied (1) iii. reserved (2) iv. address recognized (1) vi. frame copied (1) vii. reserved (2) token frame: starting delimiter (1), access control (1), ending delimiter (1) e. Token ring operation Diagram and sample data flow drawing Sending a frame from one station to another - what happens? f. Ring maintenance 1. Active monitor issues active monitor present frame every 7 seconds 2. AM detects and re-issues lost tokens 3. Stops circulating frames (monitor present bit) 4. Stops non-zero priority frames from circulating 5. Detects duplicate AMs and goes into standby mode Election process. In the absence of a active monitor, all stations (standby monitors by default) negotiate to become the active monitor (the lowest MAC address value). All stations capable of being the active monitor. - Beacon - Purge - Soft/hard errors - Fault domain (ring map) g. Ring performance As load increases, everyone is still able to transmit, but at a nearly equal degraded rate. Ring can remain operational even at very high loads (99% or higher). Ethernet cannot. h. Early token release (ETR) In theory, a token can be released immediately after the end of a normal frame transmission, so as not to waste ring capacity. In practice, the ring size (16Mb/s only) has to be very large for ETR to make any difference. For more, see Stallings LAN & MAN book.