1	Introduction to Networking
2	Introduction A distributed systems consists of a collection of computers linked together by a network Distributed system software enables computers to coordinate activities and share resources (software, hardware, data)
3	Internetworking Network: a set of computers and peripherals (printers, modems, etc.) connected together by a medium (direct or indirect connection). Network architecture: Devices in same room or building (LAN) vs. WAN
4	Communication Protocols To communicate any two entities requires establishing certain rules that can be expressed by a communication protocol. syntax semantics timing and synchronization
5	Protocol Stacks The protocol stack separates the communicating application from aspects of communication that are irrelevant to that application. The International Standardization Organization (ISO) proposed the Open System Interconnection (OSI) communication model. The Internet does NOT strictly follow the OSI model!
6	Protocol Stacks (2) Each layer presents a set of services for the layer immediately above it and can use the services provided by the layer immediately below it. The services are accessed through service access points (SAP). At each layer, certain control information is attached to the message. This is called a header. data + header = protocol data unit (PDU) the size of the message passing through the protocol stack increases at each layer!
7	Protocol Stacks (3) Message headers may contain info needed for: Connection control Addressing Segmentation and re-assembly Sequencing Error recovery Flow control Multiplexing
8	Connection Control Two fundamental types of protocols: Connection-less Each PDU is handled and routed independently The message is called a datagram Many different routing algorithms (static vs adaptive) Connection oriented a virtual circuit (VC) is maintained for the duration of the communication
9	Addressing Need to indicate the destination of the message Other info: Addressing level (e.g. IP address) Mode Unicast Multicast Broadcast
10	Segmentation and Re-assembly Long messages are cut into smaller chunks Header indicates chunk number
11	Sequencing Must maintain chronological sequence of messages to conform to protocol! Header contains sequence number
12	Error Recovery Can use sequence number to detect loss of message Request for re-send Buffering within “window” of opportunity to limit re-send requests
13	Flow Control Producer might be too fast for consumer! Need to tell producer to stop or slow down More generally: congestion control
14	The OSI Reference Model A protocol stack with seven layer
15	Internet vs OSI Driven by Internet Engineering Task Force (IETF) IETF issues Requests for Comments (RFCs) The Internet addresses layers 3 and up Layers 5, 6 and 7 are collapsed into one Application Layer Network Layer: the Internet Protocol (IP) Routing: RIP, BGP, OSPF… Transport Layer: TCP/IP and UDP/IP Application layer: ftp, smtp, http, snmp… Does the number of layers matter?
16	Finding Things: Names/Addresses To make datagram delivery possible, each computer on the net is assigned a unique identifier (well, at least one…) How many “names” exist for a “PC” How to address the webserver that manages the course webpage?
17	Addresses/Names Network Interface Hardware/MAC address, often 48-bit IEEE address (but not always: Bluetooth has 3 bit address dynamically assigned, dedicated links do not need hardware addresses) Why not use IP address? Hosts (well, technically interface as well) IP address (i.e. 134.117.63.134) DNS name/symbolic name (i.e., kunz-pc.sce.carleton.ca) Applications Port Numbers
18	More on Naming Names/Addresses better be unique (sort of) ICANN/IANA, IEEE Names/Addresses have structures, are NOT opaque identifiers, structure is exploited in locating entities IP addresses are 32-bit numeric identifiers containing network and host identifiers (see later) DNS/symbolic names have a hierarchical structure, based on well-known “top-level” domains IEEE MAC addresses: 24 bits identify manufacturer, 24 bits supposedly running number uniquely identifying each NIC built by manufacturer Port numbers: ?
19	Internet Addresses
20	Internet Addresses Different classes meet the requirements of different organizations Class A for large networks (>65,536 hosts) Class B for med-size networks (256<size<65,536) Class C for networks with up to 256 hosts Class D is used for multicasts Class E for future use Network part of Internet address is assigned by ICANN Host part is assigned by owner of the network
21	Subnet mask Allows further subdivision of network Is achieved by making special use of host number: <network number><subnet number><host number> The Subnet Mask tells you how many bits are assigned to the subnet number
22	Which Addresses/Names do End Users need to know? IP address: numerical, hard to remember Symbolic name often easier to remember Translation from Symbolic name to IP address: Domain Name Service (DNS) MAC Address: only really used/needed for MAC protocol – automatic translation from/to IP address using ARP/RARP Port Numbers: ?
23	/etc/host and /etc/services Type: cat /etc/host (or /etc/services) in a Unix system What do you see?