1 / 25

The Network Layer

The Network Layer. application layer. FTP. HTTP. SMTP. DNS. Finger. transport layer. TCP. UDP. network layer. IP. data link layer. Ethernet. ATM. modem. SHRIMP. Internetworking. challenge: how to tie together diverse networks into a single meta-network goes beyond bridges

beau
Download Presentation

The Network Layer

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The Network Layer application layer FTP HTTP SMTP DNS Finger transport layer TCP UDP network layer IP data link layer Ethernet ATM modem SHRIMP

  2. Internetworking • challenge: how to tie together diverse networks into a single meta-network • goes beyond bridges • spans heterogeneous networks • must scale to millions of hosts • terminology • internetwork: any combination of networks • Internet: a particular internetwork we all use

  3. Internetworking • three facets: • addressing: how hosts are named • service model: how hosts interact with the network • routing: how a route from sender to destination is chosen • several versions of IP exist • everybody uses version 4 now • version 6 is coming

  4. Who Runs the Internet? • nobody! • the Internet Engineering Task Force (IETF) declares technical standards, and then everybody follows them • the IETF negotiates a standard with the major players, and people do what they want, influenced by the IETF standard • who elects the IETF?

  5. 7 24 class A 0 network host 14 16 class B 10 network host class C 110 network host Addressing • unique 32-bit address for each host • two-level address: (network, host) • dotted-decimal notation (128.112.153.27) • three address formats 21 8

  6. Two-Level Addressing • network is a single (extended) LAN • provides physical address for each host • can deliver packets between all hosts in network • outside hosts treat a network as a black box • advantages of two-level approach • can remember single route for a whole network • host numbers can be allocated locally

  7. Three-Level Addressing • (network, subnet, host) • subnet and host numbers managed locally • outside can’t tell two-level from three-level • variable division of address bits between subnet and host parts • determined by subnet mask • subnet is a single LAN • subnets treat each other as black boxes

  8. Translating Addresses • IP uses IP addresses, but a network uses physical addresses • must translate to communicate within network • each host keeps a list of IP-to-physical translations • forgotten every 15 minutes or so • Address Resolution Protocol (ARP) translates IP address to physical address

  9. ARP • querying host broadcasts a packet asking for a translation for some IP address • host with that IP address answers with ARP response containing the physical address • all hosts know their own addresses • query contains translation for the querying host • other hosts learn from the query

  10. RARP • reverse Address Resolution Protocol (RARP) translates physical to IP • works just like ARP • allows a host to learn its own IP address after it boots • table of address mappings stored on a server • maintained by administrator

  11. DHCP • Dynamic Host Configuration Protocol (DHCP) allows host to learn its own network parameters • on startup, host broadcasts DHCP query • server responds with all the information host needs to function on the Internet • used on many PC networks

  12. Service Model • unreliable datagram service • transmits data in variable size chunks (datagrams) • may drop, reorder, or duplicate datagrams • best effort approach

  13. Fragmentation and Reassembly • each network has a Maximum Transmission Unit (MTU): the largest IP packet it can carry • packet may travel through diverse networks with different MTUs • solution: if packet is bigger than MTU, break it into fragments • may be broken more finely in a later network • reassemble at ultimate destination

  14. 4000 1500 1500 1000 476 512 512 476 512 512 488 512 512 512 476 512 512 476 512 488 Fragmentation Example FDDI (MTU=4500) Ethernet (MTU=1500) P2P (MTU=512) FDDI (MTU=4500)

  15. Reassembly • performed by destination host • store fragments in memory until they all show up • fragments might be dropped, so discard stored fragments after a timeout period • for best performance, avoid fragmentation • higher-level protocols can avoid sending large datagrams (use common MTU, or discover the right size dynamically)

  16. IP Packet Format 32 bits type of service total length version hdr len datagram identifier flags offset in datagram protocol checksum time to live source address destination address options (if any; variable size) data follows

  17. IP Packet Format: Details • version: currently 4 • header length: in 32-bit words, 5 if no options • type of service: almost never used • total length: in bytes, header+data • datagram identifier: allows destination to match up fragments of the same datagram

  18. IP Packet Format: Details • flags • more-fragments: says this isn’t the last fragment of the datagram • don’t-fragment: prohibits fragmentation; packet will be dropped rather than fragmented • offset: offset within datagram at which this fragment begins

  19. IP Packet Format: Details • time to live: initially set to 64; decremented on each hop; packet dropped if TTL==0 • protocol: identifies which higher-level protocol this datagram belongs to • checksum: 16-bit ones-complement sum • source address, destination address obvious • options: rarely used

  20. ICMP • Internet Control Message Protocol • companion protocol to IP • used to report errors in IP • types of packets • destination host unreachable • reassembly failed • checksum didn’t match • don’t send so fast • there’s a better route

  21. Host Names and DNS • Domain Naming System (DNS) provides a mapping from human-readable names (idea.cs.princeton.edu) to IP addresses (128.112.153.27) • benefits • easier to remember • level of indirection allows reconfiguration without telling clients

  22. Domains • hierarchy of domains • top-level: .edu .gov .uk .su and so on • deeper levels: princeton.edu cs.princeton.edu • name-servers provide mappings for a domain • at least two name servers per domain • servers for a domain must know IP addresses of name-servers for all sub-domains

  23. DNS Naming • top-level servers know IP addresses of name-servers for top-level domains • lookup process • ask top-level server for IP address for desired name • result is either • the answer • referral to name-server for a more specific domain • iterate until you have the answer

  24. DNS Details • most machines just ask a local DNS server, which carries out the lookup process • cache results • forget after timeout period • also cache lookup failures • locations of top-level servers hard-wired into local DNS servers • location of local DNS server hard-wired into each machine

  25. Administrative Details • to allocate a DNS domain name • contact InterNIC (Network Information Center) • pay them a fee • they check that you have the necessary servers running • they tell top-level servers that you’re there • similar for IP network number allocation

More Related