Presentation on theme: "Lecture 6: Internetworking Principles. Part 1 – Internetworking: The term internetworking describes the connecting of separate networks possibly based."— Presentation transcript:
Lecture 6: Internetworking Principles
Part 1 – Internetworking: The term internetworking describes the connecting of separate networks possibly based on different networking technologies and possibly belonging to different organizations together. We will begin by qualifying what is required to support this capability.
Requirements for Internetworking: Homogeneous addressing scheme that uniquely identifies all hosts regardless of location or subnet Homogeneous format for all packets transmitted and standards for handling them Equipment to interconnect heterogeneous network technologies and handle the directing of packets exchanged between the technologies towards their destinations Interconnecting Equipment
Part 2 – Internetworking Equipment: Many pieces of standard networking equipment and networking strategies have been developed to support the requirements outlined above. We will now name and describe each, and give some examples of where its use would be applicable. The layered approach to networking described earlier gives rise to our ability to mix and match varying network technologies this way in an internetwork.
Repeaters and Hubs: Physical expansion/extension of network Does NOT create a logical extension - i.e. same subnet Joins multiple shorter segments to form a larger segment Could possibly involve a change of media Will not involve a change of network protocol Hubs and repeaters detect an incoming signal and retransmit it for the primary purpose of amplifying a degraded signal, and for fanning out i.e. star configuration.
Without hubs, only two machines could communicate over twisted pair ethernet … workstation server
… and without repeaters, thin net coaxial ethernets would be restricted to a maximum of thirty nodes and less than 200 meters.
This is considered a single ethernet segment. A transmission from any one host is broadcast to all others Despite the existence of seven individual ethernet cables, this is considered a single ethernet segment. Twisted Pair Ethernet Hub workstation server
This is still a single ethernet segment. HUB
… and so is this … Repeater
Despite possible media changes by a repeater, there is still only one ethernet segment (i.e. one subnet) in this example: The signal encoding method and the format of a packet are the same for all three types of ethernet present below Repeaters may have general ethernet AUI (Attachment Unit Interfaces) which may accommodate a variety of ethernet transceivers for different media types. multiport repeater AUI1 AUI2 AUI3 coax twisted pair fiber
In all of the above: ONE ethernet segment ONE logical network ONE subnet All transmissions sent by ANY host on these example configurations would be received by all of the other hosts No routing functions are performed i.e. there are no decisions made by a hub or a repeater concerning where to send a particular packet.
Switches: A switch makes routing decisions but is not considered a router. Switches do not route higher layer protocols in the OSI model. They only deals with the packets at the Data Link Layer. Routing decisions involve sending low level packets from sender to receiver and in the typical case sender and receiver are located on two segments which connect directly to the switch. Switches are very fast, but do have to look at several bytes at the beginning of each packet. Transmissions are not generally broadcast, but restricted to the segments of the ethernet where the receiver and transmitter exist. Still, switch connected segments form a single subnet.
Assume the power is just turned on …i) A workstation sends a packet looking for the serverii) Not knowing server location, switch sends everywhereiii) The server responds … the switch notes its locationiv) The switch knows where the response goes and sends itv) All further requests and replies use appropriate ports Ethernet switch Port 1 R x D T x D Port 2 R x D T x D Port 3 R x D T x D Port 4 R x D T x D Note: This is a multi- frame animated slide. The printed copy will only show the final frame.
Bridges and Routers: These are closely related. Bridges often perform routing functions. Bridges are sometimes called Bridge/Routers. A bridge spans two different network technologies. A bridge may connect two similar technologies over a different technology. If the similar technologies are assigned to be parts of the same logical network, i.e. same subnet, then the bridge is not performing any routing functions. A router may or may not connect different technologies, but in either event, connects different subnets together. Therefore routing decisions will have to be made.
Bridge Example: Wireless Ethernet I Ethernet Bridge Note: Both sides of the bridge are extensions of the same ethernet network, Ethernet I. All traffic is broadcast back and forth across the wireless link to maintain one homogeneous ethernet subnet. Participants do not perceive the existence of a wireless link
Router Example: Ethernet I Ethernet II Ethernet Router Note: Each side of the router is a separate ethernet network. Ethernet I is on one side and Ethernet II is on the other side. Packets meant for destinations on the originating side do not cross the router.
Bridge/Router Example: Wireless Ethernet I Ethernet II Ethernet Bridge Router Here the Bridge/Routers only pass traffic across the wireless link when the source and destinations are on opposite sides of the link.
Tunnels: A tunnel allows us to run a protocol through a foreign protocol by taking an encapsulted message from the first protocol, and making it look like a message to be encasulated in the second protocol. Internet Novell Ethernet Novel Tunnel over TCP/IP Novell Netware is not traditionally routable over the internet, however tunneling makes this possible by encapsulating novell packets inside of TCP/IP packets. Novell Ethernet Novel Tunnel over TCP/IP TCP/IP = Transmission Control Protocol / Internet Protocol
TCP/IP Message Novell Header Novell Message TCP/IP Header Novell Header Novell Message A Novell Packet A TCP/IP Packet Note: This is a multi- frame animated slide. The printed copy will only show the final frame. We pretend our entire Novell packet is just a message and embed it inside a TCP/IP Packet as if it were a TCP/IP message.
TCP/IP Message Novell Header Novell Message TCP/IP Header A Novell Packet A TCP/IP Packet Note: This is a multi- frame animated slide. The printed copy will only show the final frame. At the opposite end of the tunnel, we unpack the novel packet and present it to the remote Novel Network.
Part 3 – Internet Addressing: Although it is conceivable that some other standard could be developed to internetwork different technologies and organizations together, the world has settled on a particular scheme using a network layer protocol called IP. This stands for Internet Protocol. We will begin our study of IP by considering how hosts are addressed using this protocol.
IP addresses: 32-bit number divided into four octets dotted decimal notation expresses each octet in decimal notation and separates the octets with a period. i.e NNN.NNN.NNN.NNN where NNN is an integer from 0 to 255. The first octet determines the class of the IP number and as a result the interpretation of the remaining bits. Based on the class, some bits will represent a particular network, while others will represent a particular host on that network.
The Three Primary IP Classes: Class B Class C Class A 0NNNNNNNHHHHHHHHHHHHHHHHHHHHHHHH NNNNNNNNNNNNNN NNNNNNNNNNNNNNNNNNNNN HHHHHHHHHHHHHHHH HHHHHHHH Network bitsHost bitsClass bits
Multicast: Class D 11 0NNNNNNNNNNNNNNNNNNNNHHHHHHHH Network bitsHost bitsClass bits 1 Reserved: Class E
Part 4 – Internet Protocols: We will now turn our attention to the study of the protocols used in, and the issues related to internetworking. A networking course would provide a more thorough coverage of this material. Our goal is only to understand the particular aspects of the protocols that give rise to issues more directly related to the design of distributed system models.
Internet Protocols: ARP - Address Resolution Protocol UDP - User Datagram Protocol TCP - Transmission Control Protocol
ARP: Used to associate together (i.e. bind) the internet (IP) address to any addressing scheme used at the previous layer - ex. Ethernet running at the DLL will have ethernet MAC addresses like AB:CD:EF:12:34:56 which need to be mapped to IP addresses like at the Network Layer of the OSI model. RARP is Reverse Address Resolution Protocol. - works in the opposite direction ARP: converts IP to MAC RARP: converts MAC to IP
UDP: Provides a connectionless service over IP Has no session or transport layer Talks directly to the network layer (IP) Allows messages to be sent from client to server with no guaranteed delivery and without any acknowledgement of receipt by the recipient.
TCP: Provides a connection oriented service over IP Fits into the transport and session layers of the OSI networks model Talks to the network layer (IP) Allows a client and server process to establish a virtual circuit between them which they can use as a bi-directional communications channel with guaranteed error free delivery.
Part 5 - Internet Routing: Earlier, we discussed the idea of routing messages correctly from their source to their destination in a network. We will now look at how this process is managed in IP specifically. The related term RIP will be reviewed and the term CIDR will be introduced and explained.
Routing: At the network layer routing is a non-issue - IP packets are delivered directly from host to host if they are on the same network. If the destination host is on a different network (subnet), the sender will send the packet to the local router (gateway) for routing. RIP (Router Information Protocol) keeps all such routers updated regarding paths and congestion towards the destination. Default Route: Only routes to known networks are specifically held by each router. Packets destined for other destinations will be sent towards the nearest backbone via a default route.
CIDR: Classless Internet Domain Routing in the past, the network and host bits were defined strictly on the basis of the class of the IP address, and routing could only take place on that basis Although local sub-netting was possible by use of a subnet mask used to redefine host bits as network bits, this information could not be made widely available to routers Two changes occur in CIDR - routers are aware of netmasks and subnetting - netmasks can not only redefine host bits as network bits, but can also redefine network bits as host bits.
Class C 110 NNNNNNNNNNNNNNNNNNNNNHHHHHHHH Network bitsHost bitsClass bits Prior to CIDR, a router could only view the network portion of an address as defined by its class. A local netmask could be used on the local side only to split the network up into subnets. In this example, we have eight subnets. Each network now has 32 possible hosts on it. External routers were unaware of the split. Netmask Effect 110 NNNNNNNNNNNNNNNNNNNNNNNNHHHHH
Class C 110 NNNNNNNNNNNNNNNNNNNNNHHHHHHHH Network bitsHost bitsClass bits With CIDR, an external router is aware of the netmask and can now route packets for different subnets to entirely differnet destinations. Furthermore, the netmask bits can now extend either way to form not only subnets, but supernets. In this example we have combined four networks into one larger one. Netmask Effect now recognized externally 110 NNNNNNNNNNNNNNNNNNN HHHHHHHHHH
Part 6 - IPv6: Currently, version 4 of the IP protocol is predominantly being used in the Internet. IPv6 is a new implementation referred to as IP version 6. We will discuss the benefits that this new version of IP will bring once it has been fully implemented.
IPv6: Address space is expanded from 32 bits to 128 bits No checksums since integrity can be handled elsewhere No fragmentation Support of real-time and special services Introduction of anycast mode (at least one of a group) Support of authentication and encryption at the network layer