CPS 110 Networks Landon Cox March 25 2009

CPS 110: Networks Landon Cox March 25, 2009

Network hardware reality ê Lots of different network interface cards (NICs) ê 3 Com/Intel, Ethernet/802. 11 x ê Each NIC has a fixed hardware address ê MAC address: 01: 10: C 6: CE: 8 E: 42 ê ê Send packet to LAN by specifying MAC address Max packet size is 1500 bytes Packets can be reordered, corrupted, dropped Anyone can sniff packets from the network

Virtual/physical interfaces Applications OS Hardware

Distributed computing ê Try to make multiple computers look like one ê We won’t really cover ê Take CPS 214 ê Distributed shared memory ê Distributed file systems ê Parallelizing compilers ê Process migration

Protocol layers NFS (files) HTTP (web) SMTP (email) SSH (login) RPC Applications Abstraction UDP Abstraction TCP IP Ethernet ATM Abstraction PPP Hardware

OSI model ê Open Systems Interconnections Layer 7 Applications Layer 6 Presentation Layer 5 Session Layer 4 Transport Layer 3 Network Layer 2 Layer 1 Data. Link Physical

Network layers (the stack) ê Build higher-level services on simpler ones ê IP over Ethernet ê TCP over IP ê HTTP over TCP ê Why build in layers? ê Could have 0 layers (build directly on top of HW) ê What would happen? ê Have to build from scratch each time HW changes ê E. g. one firefox for wired NIC, one for wireless NIC

Network layers (the stack) ê Build higher-level services on simpler ones ê IP over Ethernet ê TCP over IP ê HTTP over TCP ê Why build in layers? ê Could have 1 layer (OS provides single layer) ê What would happen? ê Better to let applications choose functionality they need ê Unneeded features usually cost something (performance) ê E. g. would you ever not need reliable communication?

Virtual/physical interfaces Applications OS Hardware

Routing ê HW lets us send to neighbor on same LAN ê Single-hop route ê Want to send to computer on another LAN ê Multi-hop route ê IP (Internet Protocol) handles this

Local-area network ê Typically, switched Ethernet switch ê Messages delivered using ê Ethernet MAC address ê E. g. 00: 0 D: 56: 1 E: AD: BB ê Unique to physical card (like a serial number) ê Switch knows all connected computers’ MAC addresses

Routing ê Can’t put all computers on one switch! ê Think of the wiring logistics ê Want to connect two LANs together ê Use a machine that straddles two networks ê Called a router or gateway or bridge ê LANs and routers form the Internet

Internet graph A B Each letter is a router, possibly with a LAN connected to it. C E D G F

Internet graph Each node is an Autonomous System (AS). Can think of as an ISP.

Internet graph A B C E D G F How does D know how to get to router G? Should it send messages to E, C, or F?

Internet routing is imprecise ê Internet has no centralized state ê Makes it (supposedly) more fault-tolerant ê Routing is hard when a network is ê ê Large (a lot to track) Dynamic (connections change quickly) Incentives to lie (make money by accepting traffic) The Internet exhibits all three ê Basic idea ê Routers propagate info about the graph to each other ê BGP (Border Gateway Protocol)

Traceroute example ê www. kernel. org ê Unix traceroute utility

Virtual/physical interfaces Applications OS Hardware

Naming other computers ê Low-level interface ê Provide the destination MAC address ê 00: 13: 20: 2 E: 1 B: ED ê Middle-level interface ê Provide the destination IP address ê 152. 3. 140. 183 ê High-level interface ê Provide the destination hostname ê crocus. cs. duke. edu

Translating hostname to IP addr ê Hostname IP address ê Performed by Domain Name Service (DNS) ê Used to be a central server ê /etc/hosts at SRI ê What’s wrong with this approach? ê Doesn’t scale to the global Internet

DNS ê Centralized naming doesn’t scale ê Server has to learn about all changes ê Server has to answer all lookups ê Instead, split up data ê Use a hierarchical database ê Hierarchy allows local management of changes ê Hierarchy spreads lookup work across many computers

Example: www. cs. duke. edu ê nslookup in interactive mode

Translating IP to MAC addrs ê IP address MAC address ê Performed by ARP protocol ê Only done after you get to the right LAN ê How does a router know the MAC address of 152. 3. 140. 183? ê ê ê ARP (Address Resolution Protocol) If it doesn’t know the mapping, broadcast through switch “Whoever has this IP address, please tell me your MAC address” Cache the mapping “/sbin/arp” ê Why is broadcasting over a LAN ok? ê Number of computers connected to a switch is relatively small

Virtual/physical interfaces Applications OS Hardware

Message sizes ê Hardware interface ê Max Ethernet message size is 1500 bytes ê Application interface ê IP maximum packet size is 64 kbytes ê What if the route narrows? ê Start at Ethernet max of 1500 bytes ê Could traverse ATM w/ max of 53 bytes

Message sizes ê IP layer fragments larger MTU to smaller MTU Computer 1 Router Computer 2 IP IP IP Ethernet ATM

Virtual/physical interfaces Applications OS Hardware

Processes vs machines ê IP is machine-to-machine ê E. g. crocus. cs. duke. edu www. kernel. org ê Process abstraction ê Each app thinks it has its own machine ê Give each process multiple virtual NICs

Processes vs machines ê Hardware interface ê One network endpoint per machine ê Application interface ê Multiple network endpoints per machine ê Sockets ê Software endpoints for communication ê Like virtual network cards

Sockets ê Another example of virtualized hardware ê Thread virtual processor ê Address space virtual memory ê Endpoint/socket virtual NIC ê NIC and socket both have unique identifiers ê NIC: MAC address ê Socket: ‹hostname, port number› ê bind () assigns a port number to a host’s socket

Sockets ê OS allows apps to program sockets ê E. g. BSD sockets ê Win. Sock has pretty much same interface ê Processes name each other via sockets ê Each message includes a destination ‹host, port› ê Tells routers which computer gets message ê Tells dst computer which process gets message

Sockets ê OS can multiplex multiple connections over one NIC ê Kinds of sockets: UDP (datagrams), TCP (ordered, reliable)

Course administration ê Project 2 due on Friday ê ê Two groups are done Many more are very close Normal office hours Thursday and Friday Use Friday discussion section to answer P 2 questions ê Any questions?