Question? Leave a message!




Internetworking: philosophy, addressing, forwarding, resolution, fragmentation

Internetworking: philosophy, addressing, forwarding, resolution, fragmentation 26
Internetworking: philosophy, addressing, forwarding, resolution, fragmentation Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 1Overview  Internetworking: heterogeneity scale  IP solution:  Provide new packet format and overlay it on subnets.  Ideas: Hierarchical address, address resolution, fragmentation/reassembly, packet format design, forwarding algorithm etc  Chapter 3,4,5,7 in Comer  Reading: Clark: "The Design Philosophy of the DARPA Internet Protocols":  Reading: Cerf, Kahn: "A Protocol for Packet Network Intercommunication"  Reading: Mogul etal: "Fragmentation Considered Harmful"  Reading: Addressing 101: Notes on Addressing: In PDF In MS Word  Reading: Notes for Protocol Design, E2e Principle, IP and Routing: In PDF  Reference: RFC 791: Internet Protocol (IP) Spec.: In HTML Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 2The Problem  Before Internet: different packetswitching networks (e.g., ARPANET, ARPA packet radio)  only nodes on the same network could communicate Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 3A Translationbased Solution ALG ALG ALG ALG  applicationlayer gateways  inevitable loss of some semantics  difficult to deploy new internetwide applications  hard to diagnose and remedy endtoend problems  stateful gateways inhibited dynamic routing around failures  no global addressability Shivkumar Kalyanaraman Rensselaer Polytechnic Institute  adhoc, applicationspecific solutions 4The Internetworking Problem  Two nodes communicating across a “network of networks”… How to transport packets through this heterogeneous mass A B Cloud Cloud Cloud Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 5Declared Goal  “…both economic and technical considerations lead us to prefer that the interface be as simple and reliable as possible and deal primarily with passing data between networks using different packet switching strategies” V. G. Cerf and R. E. Kahn, 1974 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 6The Challenge: Heterogeneity  Share resources of different packet switching networks  interconnect existing networks  … but, packet switching networks differ widely  different services e.g., degree of reliability  different interfaces e.g., length of the packet that can be transmitted, address format  different protocols e.g., routing protocols Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 7The Challenge: Scale  Allow universal interconnection  Mantra: Connectivity is its own reward  … but, core protocols had scalability issues  Routing algorithms were limited in the number of nodes/links they could handle and were unstable after a point  Universal addressing to go with routing  As large numbers of users are multiplexed on a shared system, a congestion control paradigm is necessary for stability  No universal, scalable naming system… Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 8The Internetworking Problem  Problems: heterogeneity and scaling  Heterogeneity: How to interconnect a large number of disparate networks (lower layers) How to support a wide variety of applications (upper layers)  Scaling: How to support a large number of endnodes and applications in this interconnected network Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 9Solution Network Layer Gateways Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 10The IP Solution … IP IP IP IP  internetlayer gateways global addresses  simple, applicationindependent, lowest denominator network service: besteffort datagrams  stateless gateways could easily route around failures  with applicationspecific knowledge out of gateways:  NSPs no longer had monopoly on new services  Internet: a platform for rapid, competitive innovation Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 11Networklayer Overlay model Define a new protocol (IP) and map all applications/networks to IP  Require only one mapping (IP new protocol) when a new protocol/app is added  Global address space can be created for universal addressibility and scaling Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 12Before IP (FTP – File Transfer Protocol, NFS – Network File Transfer, HTTP – World Wide Web protocol) FTP NFS HTTP Application Telnet Coaxial Fiber Packet Transmission cable optic radio Media  No network level overlay: each new application has to be reimplemented for every network technology Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 13IP  Key ideas:  Overlay: better than anyany translation. Fewer, simpler mappings.  Networklayer: efficient implementation, global addressing FTP NFS HTTP Telnet Application Intermediate Layer (IP) Coaxial Fiber Packet Transmission cable optic radio Media Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 14What About the Future  Internet is running out of addresses  Solutions Classless Inter Domain Routing (CIDR) Network Address Translator (NATs) Dynamic Address Assignments … IPv6  Why not variablesized addresses Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 15Service to Apps  Unbounded but finite length messages byte streaming (What are the advantages)  Reliable and insequence delivery  Full duplex  Solution: Transmission Control Protocol (TCP) Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 16Original TCP/IP (Cerf Kahn)  No separation between transport (TCP) and network (IP) layers  One common header use ports to multiplex multiple TCP connections on the same host 32 32 16 16 8n Source/Port Source/Port Window ACK Text  Bytebased sequence number (Why)  Flow control, but not congestion control Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 17Today’s TCP/IP  Separate transport (TCP) and network (IP) layer (why) split the common header in: TCP and UDP headers fragmentation reassembly done by IP  Congestion control (later in class) Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 18IP Datagram Format 0 4 8 16 32 Vers H Len TOS Total Length Identification Flags Fragment Offset Time to live Protocol Header Checksum Source IP Address Destination IP Address IP Options (if any) Padding Data Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 19IP Datagram Format (Continued)  First Word purpose: info, variable size header packet. Version (4 bits) Internet header length (4 bits): units of 32bit words. Min header is 5 words or 20 bytes. Type of service (TOS: 8 bits): Reliability, precedence, delay, and throughput. Not widely supported Total length (16 bits): header + data. Units of bytes. Total must be less than 64 kB. Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 20IP Header (Continued)  2nd Word Purpose: fragmentation Identifier (16 bits): Helps uniquely identify the datagram between any source, destination address Flags (3 bits): More Flag (MF):more fragments Don’t Fragment (DF) Reserved Fragment offset (13 bits): In units of 8 bytes Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 21IP Header (Continued)  Third word purpose: demuxing, error/looping control, timeout. Time to live (8 bits): Specified in router hops Protocol (8 bits): Next level protocol to receive the data: for demultiplexing. Header checksum (16 bits): 1’s complement sum of all 16bit words in the header.  Change header = modify checksum using 1’s complement arithmetic. Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 22Recall: Signed Representations Sign Magnitude One's Complement Two's Complement 000 = +0 000 = +0 000 = +0 001 = +1 001 = +1 001 = +1 010 = +2 010 = +2 010 = +2 011 = +3 011 = +3 011 = +3 100 = 0 100 = 3 100 = 4 101 = 1 101 = 2 101 = 3 110 = 2 110 = 1 110 = 2 111 = 3 111 = 0 111 = 1 One’s complement addition: normal addition increment of the total if there was a carry. Eg: 110 (i.e. 1) + 111 (i.e. 0) = 101 +1 = 110 (i.e. –1) Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 23Header Format (Continued)  Source Address (32 bits): Original source. Does not change along the path  Destination. Address (32 bits): Final destination. Does not change along the path.  Options (variable length): Security, source route, record route, stream id (used for voice) for reserved resources, timestamp recording  Padding (variable length): Makes header length a multiple of 4  Payload Data (variable length): Data + header 65,535 bytes Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 24TCP Header 0 4 10 16 31 Destination port Source port Sequence number Acknowledgement Advertised window HdrLen Flags Checksum Urgent pointer Options (variable)  Sequence number, acknowledgement, and advertised window – used by slidingwindow based flow control  Flags (selected): SYN, FIN – establishing/terminating a TCP connection ACK – set when Acknowledgement field is valid RESET – abort connection Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 25TCP Header (Cont)  Checksum – 1’s complement and is computed over TCP header TCP data Pseudoheader (from IP header) Note: breaks the layering Source address Destination address TCP Segment length 0 Protocol (TCP) Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 26TCP Connection Establishment  Threeway handshake Goal: agree on a set of parameters: the start sequence number for each side Server Client (initiator) Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 27IP Forwarding (I)  Source Destination in same network (direct connectivity) Recognize that destination IP address is on 1 same network. 2 Find the destination LAN address. Send IP packet encapsulated in LAN frame directly to the destination LAN address.  Encapsulation = source/destination IP addresses don’t change Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 28IP Forwarding (II)  B) Source Destination in different networks (indirect connectivity) Recognize that destination IP address is not 1 on same network. Look up destination IP address in a (L3 forwarding) table to find a match, called the next hop router IP address. Send packet encapsulated in a LAN frame to the LAN address corresponding to the IP 2 address of the nexthop router. Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 291 Addressing  1 How to find if destination is in the same network IP address = network ID + host ID.  Source and destination network IDs match = same network (I.e. direct connectivity) Splitting address into multiple parts is called hierarchical addressing Network Host Boundary Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 302 Address Resolution  2: How to find the LAN address corresponding to an IP address Address Resolution Problem. Solution: ARP, RARP (later in this slide set) Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 31IP Forwarding: Example Scenario routing table in A Dest. Net. next router Nhops 223.1.1 1 223.1.2 223.1.1.4 2 IP datagram: 223.1.3 223.1.1.4 2 source misc dest data IP addr fields IP addr A 223.1.1.1 datagram remains 223.1.2.1 223.1.1.2 unchanged, as it travels 223.1.2.9 223.1.1.4 source to destination B addr fields of interest here 223.1.2.2 223.1.3.27 E 223.1.1.3 223.1.3.2 223.1.3.1 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 32IP Forwarding (Direct) misc Dest. Net. next router Nhops data 223.1.1.1 223.1.1.3 fields 223.1.1 1 223.1.2 223.1.1.4 2 Starting at A, given IP 223.1.3 223.1.1.4 2 datagram addressed to B: look up net. address of B A 223.1.1.1 find B is on same net. as A 223.1.2.1 link layer will send datagram 223.1.1.2 directly to B inside linklayer 223.1.2.9 223.1.1.4 frame B B and A are directly 223.1.2.2 E 223.1.3.27 223.1.1.3 connected 223.1.3.2 223.1.3.1 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 33IP Forwarding (Indirect): Step 1 misc Dest. Net. next router Nhops 223.1.1.1 223.1.2.2 data fields 223.1.1 1 223.1.2 223.1.1.4 2 Starting at A, dest. E: 223.1.3 223.1.1.4 2 look up network address of E E on different network A 223.1.1.1 A, E not directly 223.1.2.1 attached 223.1.1.2 routing table: next hop router to 223.1.2.9 223.1.1.4 E is 223.1.1.4 B link layer sends datagram to 223.1.2.2 E 223.1.3.27 223.1.1.3 router 223.1.1.4 inside linklayer frame 223.1.3.2 223.1.3.1 datagram arrives at 223.1.1.4 continued….. Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 34IP Forwarding (Indirect): Step 2 Dest. next misc network router Nhops interface 223.1.1.1 223.1.2.2 data fields 223.1.1 1 223.1.1.4 223.1.2 1 223.1.2.9 Arriving at 223.1.4, 223.1.3 1 223.1.3.27 destined for 223.1.2.2 look up network address of E A 223.1.1.1 E on same network as router’s 223.1.2.1 interface 223.1.2.9 223.1.1.2 router, E directly 223.1.2.9 223.1.1.4 B attached 223.1.2.2 link layer sends datagram to E 223.1.3.27 223.1.1.3 223.1.2.2 inside linklayer frame 223.1.3.2 223.1.3.1 via interface 223.1.2.9 datagram arrives at 223.1.2.2 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 35The Internet Network layer Host, router network layer functions: Transport layer: TCP, UDP IP protocol Routing protocols •addressing conventions •path selection •datagram format •RIP, OSPF, BGP Network •packet handling conventions layer routing ICMP protocol table •error reporting •router “signaling” Link layer physical layer Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 36IP Addressing: introduction  IP address: 32bit identifier 223.1.1.1 for host, router interface 223.1.2.1  Interface: connection 223.1.1.2 between host, router and 223.1.2.9 223.1.1.4 physical link 223.1.2.2  router’s typically have 223.1.3.27 223.1.1.3 multiple interfaces  host may have multiple interfaces 223.1.3.2 223.1.3.1  IP addresses associated with interface, not host, router 223.1.1.1 = 11011111 00000001 00000001 00000001  Hosts in the same network 223 1 1 1 have same network ID Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 37IP Address Formats 0 Network Host  Class A: 1 7 24 bits 10 Network Host  Class B: 2 14 16 bits 110 Network Host  Class C: 3 21 8 bits 1110 Multicast Group addresses  Class D: 4 28 bits  Class E: Reserved. Router Router Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 38Dotted Decimal Notation  Binary: 11000000 00000101 00110000 00000011 Hex Colon: C0:05:30:03 Dotted Decimal: 192.5.48.3 Class Range A 0 through 127 B 128 through 191 C 192 through 223 D 224 through 239 E 240 through 255 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 39Subnet Addressing  Classful addressing inefficient: Everyone wants class B addresses  Can we split class A, B addresses spaces and accommodate more networks Need another level of hierarchy. Defined by “subnet mask”, which in general specifies the sets of bits belonging to the network address and host address respectively Network Host Boundary is flexible, and defined by subnet mask Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 40Understanding Prefixes and Masks 12.5.9.16 is covered by prefix 12.4.0.0/15 12.5.9.16 00001100 00000101 00001001 00010000 00001100 00000100 00000000 00000000 12.4.0.0/15 11111111 11111110 00000000 00000000 12.7.9.16 00001100 00000111 00001001 00010000 12.7.9.16 is not covered by prefix 12.4.0.0/15 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 41RFC 1519: Classless InterDomain Routing (CIDR) PreCIDR: Network ID ended on 8, 16, 24 bit boundary CIDR: Network ID can end at any bit boundary IP Address : 12.4.0.0 IP Mask: 255.254.0.0 Address 00001100 00000100 00000000 00000000 Mask 11111111 11111110 00000000 00000000 Network Prefix for hosts Usually written as 12.4.0.0/15, a.k.a “supernetting” Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 42Interdomain Routing Without CIDR 204.71.0.0 204.71.0.0 204.71.1.0 Global 204.71.1.0 204.71.2.0 Service Internet 204.71.2.0 …...……. Provider Routing …...……. Mesh 204.71.255.0 204.71.255.0 Interdomain Routing With CIDR 204.71.0.0 204.71.1.0 Global 204.71.2.0 Service Internet 204.71.0.0/16 …...……. Provider Routing Mesh 204.71.255.0 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 43Implication on Forwarding: Subnet Route table lookup:  IF ((Maski Destination Addr) = = Destinationi) Forward to NextHopi In theory, subnet mask can end on any bit. In practice, mask must have contiguous 1s followed by contiguous zeros. Routers do not support other types of masks. So, (Address, Mask) = (12.4.0.0, 255.254.0.0) may be written as 12.4.0.0/15 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 44Route Table Lookup: Subnet Example 30.0.0.7 40.0.0.8 128.1.0.9 40.0.0.0 30.0.0.0 128.1.0.0 192.4.0.0 40.0.0.7 128.1.0.8 192.4.10.9 Destination Mask Next Hop 30.0.0.0 255.0.0.0 40.0.0.7 40.0.0.0 255.0.0.0 Deliver direct 128.1.0.0 255.255.0.0 Deliver direct 192.4.10.0 255.255.255.0 128.1.0.9 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 45Implication on Forwarding: Supernetting (CIDR) • Longest Prefix Match (Classless) Forwarding Destination =12.5.9.16 payload Prefix Next Hop Interface OK 0.0.0.0/0 10.14.11.33 ATM 5/0/9 better 12.0.0.0/8 10.14.22.19 ATM 5/0/8 12.4.0.0/15 10.1.3.77 Ethernet 0/1/3 even better best 12.5.8.0/23 attached Serial 1/0/7 IP Forwarding Table Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 46Variable Length Subnet Mask (VLSM)  Basic subneting: refers to a fixed mask in addition to natural mask (i.e. class A, B etc).  I.e. only a single mask (eg:: 255.255.255.0) can be used for all networks covered by the natural mask.  VLSM: Multiple different masks possible in a single class address space.  Eg: 255.255.255.0 and 255.255.254.0 could be used to subnet a single class B address space.  Allows more efficient use of address space. Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 47Example: Address Block: 128.20.224.0/20. Networks: 2 of size 1000 nodes each; 2 of size 500 nodes each; 3 of size 250 nodes each. 4 of size 50 nodes each. What are the allocations 1000 nodes need 10 bits = 32 –10 =22 bit prefixes needed 128.20.1110 00 00. 0000 0000/22 = 128.20.224.0/22 128.20.1110 01 00. 0000 0000/22 = 128.20.228.0/22 500 nodes need 9 bits = 32 –9 =23 bit prefixes needed 128.20.1110100 0. 0000 0000/23 = 128.20.232.0/23 128.20.1110101 0. 0000 0000/23 = 128.20.234.0/23 250 nodes need 8 bits = 32 –8 =24 bit prefixes needed 128.20.11101100. 0000 0000/24 = 128.20.236.0/24 128.20.11101101. 0000 0000/24 = 128.20.237.0/24 128.20.11101110. 0000 0000/24 = 128.20.238.0/24 50 nodes need 6 bits = 32 –6 =26 bit prefixes needed Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 48Addressing Summary Unique IP address per interface Classful (A,B,C) = address allocation not efficient Hierarchical = smaller routing tables Provision for broadcast, multicast, loopback addresses Subnet masks allow “subnets” within a “network” = improved address allocation efficiency Supernet (CIDR) allows variable sized network ID allocation VLSM allows further efficiency Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 49Forwarding Summary  Forwarding: Simple “nexthop” forwarding. Last hop forwards directly to destination Besteffort delivery : No error reporting. Delay, outoforder, corruption, and loss possible = problem of higher layers Forwarding vs routing: tables setup by separate algorithm (s) Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 50What IP does NOT provide  Endtoend data reliability flow control (done by TCP or application layer protocols)  Sequencing of packets (like TCP)  Error detection in payload (TCP, UDP or other transport layers)  Error reporting (ICMP)  Setting up route tables (RIP, OSPF, BGP etc)  Connection setup (it is connectionless)  Address/Name resolution (ARP, RARP, DNS)  Configuration (BOOTP, DHCP)  Multicast (IGMP, MBONE) Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 51Maximum Transmission Unit  Each subnet has a maximum frame size Ethernet: 1518 bytes FDDI: 4500 bytes Token Ring: 2 to 4 kB  Transmission Unit = IP datagram (data + header)  Each subnet has a maximum IP datagram length (header + payload) = MTU Net 1 Net 2 S R R MTU=1500 MTU=1000 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 52Fragmentation  Datagrams larger than MTU are fragmented  Original header is copied to each fragment and then modified (fragment flag, fragment offset, length,...)  Some option fields are copied (see RFC 791) IP Header Original Datagram IP Hdr 1 Data 1 IP Hdr 2 Data 2 IP Hdr 3 Data 3 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 53Fragmentation Example MTU = 1500B MTU = 280B IHL = 5, ID = 111, More = 0 IHL=5, ID = 111, More = 1 Offset = 0W, Len = 472B Offset = 0W, Len = 276B IHL=5, ID = 111, More = 0 Offset = 32W, Len = 216B Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 54Fragmentation Example (Continued)  Payload size 452 bytes needs to be transmitted  across a Ethernet (MTU=1500B) and a SLIP line (MTU=280B)  Length = 472B, Header = 20B = Payload = 452B  Fragments need to be multiple of 8bytes.  Nearest multiple to 260 (280 20B) is 256B  First fragment length = 256B + 20B = 276B.  Second fragment length = (452B 256B) + 20B = 216B Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 55Reassembly  Reassembly only at the final destination  Partial datagrams are discarded after a timeout  Fragments can be further fragmented along the path. Subfragments have a format similar to fragments.  Minimum MTU along a path  Path MTU S D Net 1 Net 2 Net 3 R1 R2 MTU=1500 MTU=1000 MTU=1500 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 56Further notes on Fragmentation  Performance: single fragment lost = entire packet useless. Waste of resources all along the way. Ref: Kent Mogul, 1987  Don’t Fragment (DF) bit set = datagram discarded if need to fragment. ICMP message generated: may specify MTU (default = 0)  Used to determine Path MTU (in TCP UDP)  The transport and application layer headers do not appear in all fragments. Problem if you need to peep into those headers. Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 57Resolution Problems and Solutions  Indirection through addressing/naming = requires address/name resolution  Problem is to map destination layer N address to its layer N1 address to allow packet transmission in layer N1. Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 58Resolution Problems and Solutions (Continued)  1. Direct mapping: Make the physical addresses equal to the host ID part. Mapping is easy. Only possible if admin has power to choose both IP and physical address. Ethernet addresses come preassigned (so do part of IP addresses). Ethernet addresses are 48 bits vs IP addresses which are 32bits. Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 59ARP techniques (Continued)  2: Table Lookup: Searching or indexing to get MAC addresses Similar to lookup in /etc/hosts for names Problem: change Ethernet card = change table IP Address MAC Address 197.15.3.1 0A:4B:00:00:07:08 197.15.3.2 0B:4B:00:00:07:00 197.15.3.3 0A:5B:00:01:01:03 Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 60ARP techniques (Continued)  3. Dynamic Binding: ARP The host broadcasts a request: “What is the MAC address of 127.123.115.08” The host whose IP address is 127.123.115.08 replies back: “The MAC address for 127.123.115.08 is 8A5F3C234556 ” 16 ARP responses cached; LRU + Entry Timeout  All three methods are allowed in TCP/IP networks. Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 61ARP Message Format 0 8 16 24 32 H/W Address Type Protocol Address Type H/W Adr Len Prot Adr Len Operation Sender’s h/w address (6 bytes) Sender’s Prot Address (4 bytes) Target h/w address (6 bytes) Target Protocol Address (4 bytes)  Type: ARP handles many layer 3 and layer 2s  Protocol Address type: 0x0800 = IP  Operation: 1= Request, 2=Response  ARP messages are sent directly to MAC layer Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 62Back to Goals (Clark’88) 0 Connect existing networks  initially ARPANET and ARPA packet radio network 1. Survivability ensure communication service even in the presence of network and router failures 2. Support multiple types of services 3. Must accommodate a variety of networks 4. Allow distributed management 5. Allow host attachment with a low level of effort 6. Be cost effective 7. Allow resource accountability Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 631. Survivability  Continue to operate even in the presence of network failures (e.g., link and router failures)  as long as the network is not partitioned, two endpoint should be able to communicate…moreover, any other failure (excepting network partition) should be transparent to endpoints  Decision: maintain state only at endpoints (fate sharing)  eliminate the problem of handling state inconsistency and performing state restoration when router fails  Internet: stateless network architecture Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 642. Types of Services  Add UDP to TCP to better support other types of applications e.g., “realtime” applications  This was arguably the main reasons for separating TCP and IP  Provide datagram abstraction: lower common denominator on which other services can be built service differentiation was considered (remember ToS), but this has never happened on the large scale (Why) Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 653. Variety of Networks  Very successful (why)  because the minimalist service; it requires from underlying network only to deliver a packet with a “reasonable” probability of success  …does not require:  reliability  inorder delivery  The mantra: IP over everything  Then: ARPANET, X.25, DARPA satellite network..  Now: ATM, SONET, WDM… Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 66Other Goals  Allow distributed management  Remember that IP interconnects networks  each network can be managed by a different organization  different organizations need to interact only at the boundaries  … but this model complicates routing  Cost effective  sources of inefficiency  header overhead  retransmissions  Routing  …but routers relatively simple to implement (especially software side) Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 67Other Goals (Cont)  Low cost of attaching a new host not a strong point  higher than other architecture because the intelligence is in hosts (e.g., telephone vs. computer) bad implementations or malicious users can produce considerably harm (remember fate sharing)  Accountability very little so far Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 68What About the Future  Datagram not the best abstraction for:  resource management,accountability, QoS  A new abstraction: flow  Routers require to maintain perflow state (what is the main problem with this raised by Clark)  state management  Proposed Solution  softstate: endhosts responsible to maintain the state  Problem: increase in controltraffic to maintain state, unless efficiently piggybacked Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 69Summary: Internet Architecture  Packetswitched datagram TCP UDP network  IP is the glue (network layer overlay) IP  Hourglass architecture all hosts and routers run IP Satellite  Stateless architecture Ethernet ATM no per flow state inside network Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 70Summary: Minimalist Approach  Dumb network  IP provide minimal functionalities to support connectivity  addressing, forwarding, routing  Smart end system  transport layer or application performs more sophisticated functionalities  flow control, error control, congestion control  Advantages  accommodate heterogeneous technologies (Ethernet, modem, satellite, wireless)  support diverse applications (telnet, ftp, Web, X windows)  decentralized network administration Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 71Connect Existing Networks  Existing networks: ARPANET and ARPA packet radio  Decision: packet switching Existing networks already were using this technology  Packet switching store and forward router architecture  Internet: a packet switched communication network consisting of different networks connected by storeand forward routers Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 72Summary  Internetworking Problem  IP header: supports connectionless delivery, variable length pkts/headers/options, fragmentation/reassembly,  Fragmentation/Reassembly, Path MTU discovery.  ARP, RARP: address mapping  Internet architectural principles Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 73
sharer
Presentations
Free
Document Information
Category:
Presentations
User Name:
Dr.ShivJindal
User Type:
Teacher
Country:
India
Uploaded Date:
19-07-2017