480 likes | 619 Vues
Managing Computer and Communication Networks. Part1 Dr. Hamid Reza Naji. Motivation for Networks. Information Access Sharing of Resources Facilitate Communications. What A Network Includes. Transmission hardware Special-purpose hardware devices interconnect transmission media
E N D
Managing Computer and Communication Networks Part1 Dr. Hamid Reza Naji
Motivation for Networks • Information Access • Sharing of Resources • Facilitate Communications
What A Network Includes • Transmission hardware • Special-purpose hardware devices • interconnect transmission media • control transmission • run software protocol • Software Protocol • encodes/decodes and formats data • detects and corrects problems
What A Network Does • Provides communication that is • Reliable • Fair • Efficient • From one application to another • Automatically detects and corrects • Data corruption • Data loss • Duplication • Out-of-order delivery • Automatically finds optimal path from source to destination
Issues we’ll be dealing with Point-to-point - Individual connections between pairs of machines. Broadcast Networks - Single channel with multiple machines connected to it. Packets - Messages - the "chunk" of data transmitted from one machine to the next. Addressing - One to one: Packet contains specific target address. Broadcasting: All machines on the network receive and process the packet. Multicasting: A subset of machines receive and process the packet.
transmission distance view messaging in terms of the transmission distance 100 meter Building -- Local area network 10 kilometer City -- Metropolitan area network 1K kilometer Country -- Wide area network 10K kilometer Continents -- Internet
Dedicated fixed bandwidth route fixed at setup idle capacity wasted network state Circuit Switching
Packet Switching • Best Effort • end-to-end control • multiplexing technique • re-route capability • congestion problems
Network Hardware LOCAL AREA NETWORKS • LANs are: • Privately owned. Can be up to several kilometers long; • Run at speeds of 10, 100, or more Mbps. • Low delay. High reliability. • Requires collision arbitration. • Ethernet: • IEEE 802.3. • Bus based broadcast network with decentralized control at 10 or 100 Mbps. • Token Ring: • IEEE 802.5 • Ring based broadcast network with token arbitration at 4 or 16 Mbps.
Network Hardware METROPOLITAN AREA NETWORKS MANS are: Larger version of LAN ("city" wide). Public or private / data or voice. Broadcast Can be distinguished from LANs based on wiring mechanism.
Network Hardware WIDE AREA NETWORKS WANS are: Networks spanning large distances. Hosts or End Systems - Machines running user applications. (Communication) Subnet - Connections between hosts - transmission lines + switches. A "locality" understanding each other's addresses. Circuits/Channels/Trunks - Transmission lines move the bits.
Network Hardware WIDE AREA NETWORKS Router - Generic term for switching computers. Point-to-point/Store-and-forward/Packet-switched - Moving through a series of routers, packets are received at a router, stored there, then forwarded to the next router.
Network Hardware Media Wireline String, Copper (Twisted Pair, Coax), Optical Fiber • Wireless • Sound, Light, Infrared, RF, Microwave • Used where computer is mobile or far away from wires. • Only few Mbps / higher error rates / interference.
PROTOCOL HIERARCHIES • Layers - The concept that network software is organized functionally into levels. A level on one host talks to the same level on another host (its peer). • Protocol - The protocol is the convention or standard that a layer uses to talk to the other layer. An agreement or standard on the conversation.
PROTOCOL HIERARCHIES • Physical Medium - Underneath the layers is the wire or fiber or whatever. • Interface - Defines the services that one layer offers another (either up or down.) Important that each layer perform specific actions.
PROTOCOL HIERARCHIES • Network architecture - A set of layers and protocols. It contains details on what happens in the layer and what the layer says to its peer. • Protocol stack - A list of protocols used by a system, one protocol per layer.
DESIGN ISSUES FOR THE LAYERS • Both Directions Simultaneous • Simplex No No • Half duplex Yes No • Full duplex Yes Yes • o Number of logical channels per connection • o Error control. • o Flow control. • o Breaking up messages into a smaller chunks (and • reassembly.) • o Multiplexing messages on same connection. • o Routing - how to get from one host to another.
INTERFACES AND SERVICES • Purpose of each layer is to provide services to the layer above it. • Entities / Peer entities - • Active element in each layer (processor / IO chip). • Peer entity = layer N entity <--> layer N entity. • Layer N entity provides service for layer N + 1. • Service providers and users - • Layer N is a provider for user N + 1. • SAPs (Service Access Points) - • Entry points in N that layer N + 1 can access. • Has an address that uniquely identifies it.
INTERFACES AND SERVICES • IDUs (Interface Data Unit) - • The information from N + 1 provided at the SAP. • Made up of SDU + control information. • SDUs (Service Data Unit) - • The portion of the IDU that will be passed up to the peer entity. • PDUs (Protocol Data Unit) - • The SDU may be broken up into PDUs, that being the chunk size for further transmission.
CONNECTION-ORIENTED SERVICES • Connection oriented service - Like the phone system. The system establishes a connection, uses it, and closes it. Acts like a tube. Data comes out the other end in the same order as it goes in. • Connection Setup • Data Transfer • Connection Termination
CONNECTIONLESS SERVICES • Connectionless service - Like the post office. Each message has the entire address on it. Each message may follow a different route to its destination. Ordering is not important
CONNECTION-ORIENTED / CONNECTIONLESS SERVICES: • Quality of service - • Will the message arrive?? • A reliable connection-oriented service guarantees success. • Message sequence - message boundaries and order are maintained. • Byte streams - messages are broken up or combined; flow is bytes. • Can pair mechanism with upper-layer requirements (matching).
CONNECTION-ORIENTED / CONNECTIONLESS SERVICES: • Datagram Service – Like junk mail. It's not worth the cost to determine if it actually arrived. Needs a high probability of arrival, but 100% not required. Connectionless, no acknowledgment. • Acknowledged datagram service - As above, but improved reliability via acknowledgment. • Request-reply service - Acknowledgment is in the form of a reply.
SERVICE PRIMATIVES • Primitives are: • The operations available to an entity. Possibilities include: • Request -- An entity want some work done. • Indication -- An entity is told about an event. • Response -- An entity wants to respond to an event. • Confirm -- Response to earlier request has come back.
SERVICE PRIMATIVES • Example of Connectionless Protocol: • Data-request |--------->| Data-indication • Data-indication |<---------| Data-request • Example of Connection-oriented Protocol: (needs connection establishment) • Connect-request |--------->| Connect-indication • Connect-confirm |<---------| • Data-request |--------->| Data-indication • Data-indication |<---------| Data-request • Disconnect-request |--------->| Disconnect-indication • Disconnect-confirm |<---------|
SERVICE PRIMATIVES • THE RELATIONSHIP OF SERVICES TO PROTOCOLS: • Services are primitives that a layer provides for the layer above it. • Protocols are rules governing the meaning of frames/packets/messages exchanged with the peer entity.
Reference Models Headers, Data, and Trailers Encapsulation
THE OSI REFERENCE MODEL OSI == Open Systems Interconnection Developed by ISO == International Standards Organization Principles used to develop OSI Layering: 1. Need a layer for each different level of operation. 2. Each layer performs a well defined function. 3. Each layer has a standard. 4. Layer boundaries should minimize data flow across those boundaries. 5. The right number of layers - don't put too many functions together, but not too many layers either.
THE OSI REFERENCE MODEL • Physical Layer - • Purpose -- Transmits raw bits across a medium. • Electrical -- Concerns are voltage, timing, duplexing, connectors, etc.
THE OSI REFERENCE MODEL • Data Link Layer - • Framing -- Breaks apart messages into frames. Reassembles frames into messages. • Error handling -- solves damaged, lost, and duplicate frames. • Flow control -- keeps a fast transmitter from flooding a slow receiver. • Gaining Access -- if many hosts have usage of the medium, how is access arbitrated.
THE OSI REFERENCE MODEL • Network Layer - • Routing -- What path is followed by packets from source to destination. Can be based on a static table, when the connection is created, or dynamic when each packet is sent. • Congestion -- Controls the number packets in the subnet. • Accounting -- Counts packets/bytes for billing purposes. • Heterogeneity -- Interfacing so one type of network can talk to another.
THE OSI REFERENCE MODEL • Transport Layer - • Reliability -- Ensures that packets arrive at their destination. Reassembles out of order messages. • Hides network -- Allows details of the network to be hidden from higher level layers. • Service Decisions -- What type of service to provide; error-free point to point, datagram, etc. • Mapping -- Determines which messages belong to which connections. • Naming -- "Send to node xyz" must be translated into an internal address and route. • Flow control -- keeps a fast transmitter from flooding a slow receiver.
THE OSI REFERENCE MODEL • Session Layer – • Sessions -- Provides services that guarantee a particular message. For instance, a login session could be logged. • Synchronization-- Provide way to subdivide a long mechanism for reliability.
THE OSI REFERENCE MODEL • Presentation Layer - Prettiness -- Syntax and semantics of information transmitted. --Understands the nature of the data being transmitted. --Converts ASCII / EBCDIC.
THE OSI REFERENCE MODEL • Application Layer - • Interfacing -- Terminal type translation. • File transfer -- Programs able to understand directory structures and naming conventions and map them onto various systems.
THE TCP/IP REFERENCE MODEL Used in the Internet. Common mechanism that is surpassing the OSI Model. Internet Layer - The IP (Internet Protocol) does delivery and congestion control.
THE TCP/IP REFERENCE MODEL • Transport Layer - Allows peer entities to communicate. • TCP -- Transmission Control Protocol provides a reliable connection oriented protocol that delivers a byte stream from one node to another. Guarantees delivery and provides flow control. • UDP -- User Datagram Protocol provides an unreliable connection-less protocol for applications that provide their own. • Application Layer - Terminal -- Telnet File transfer -- FTP The Web -- HTTP
COMPARISON OF REFERENCE MODELS LAYERS Transport Network OSI Connectionless Connection-Oriented Connection-Oriented TCP Connectionless Connectionless Connection-Oriented
Example Networks • NOVELL NETWARE: Heavily used in PC world. • Proprietary protocol stack Network layer -- IPX - unreliable connectionless Transport layer -- NCP (Network Core Protocol) - connection oriented • Naming and Addressing -- • SAP (Service Advertising Protocol) - Servers advertise their address to router machines. Clients, when booted, ask for location of nearest server.
Example Networks THE INTERNET: • Growing exponentially. • All nodes run TCP/IP. Means that all nodes have an IP address by which they can be contacted. • Services provided include: o e-mail o news o remote login o file transfer o the web
Standards ISO (International Standards Organization) ANSI (American National Standards Institute) NIST (National Institute of Standards and Technology) IEEE (Institute of Electrical and Electronics Engineering) IAB (Internet Architecture Board)