Day 13 Intro to MANs and WANs

MANs Cover a larger distance than LANs –Typically multiple buildings, office park Usually in the shape of a ring –Typically multiple rings provide redundancy Typically employ SONET –OC1+ –Sometimes offers dynamically sizeable bandwidth

WAN Larger area network than MAN. Can be –Circuit Switched –Packet Switched There is also a concept of: –Virtual circuit packet switched

Packet A packet is simply a “Chunk” of data which is to be sent from one location to another. –Sometimes packets are fixed size, other times they are variable size –Transmitting a large file from one location to another is done by breaking the file into small “chunks” (packets). Each packet can then be individually error checked –If a packet is bad it doesn’t require retransmitting entire conversation.

Circuit Switched Physical circuit is maintained while connection is active. –E.g. Phone network Connection is established between 2 sides –All packets follow the same path –Connection must be dismantled at the end of the connection. Not very redundant, if connection fails, it fails. –However, no contention for resources, you own the medium –Also not very efficient, if used for data since data by its nature is sporadic Surfing the web requires only short bursts of data

Packet Switched - Datagram Each packet is routed separately. –2 packets in the same stream may follow different paths. –Decision is made at each “node” which direction to send it next. This allows for failover in the case of a link failure Allows for rerouting when lots of traffic saturates a link. Data packets are of a fixed size. Disadvantage –Takes time to decide which way to route each packet

Packet Switched – Virtual Circuit A compromise between Circuit and packet switched –Initially a “virtual circuit” is established in software. –No physical wires are changed, just routing tables in routers/switches All packets in connection are routed in this virtual circuit. –Cuts down on the overhead of packet switched, but gives its advantages.

Broadcast network E.g. Cable TV Single node sends traffic to all nodes on network No connection is established No routing done, all packets are sent to all users Everyone hears transmission –No security/privacy

Connection vs Connectionless Connection oriented (reliable) –A connection (either virtual or physical) is formed before data is transferred. It is torn down after communication is finished –Traffic can be error controlled Each packet can be acknowledged –You know if it got there correctly or not Packet errors can be detected and dealt with Connectionless (unreliable) –Each packet is on its own –Not reliable.

Routing 2 computers are communicating over the internet –Data is broken up into packets –Each packet leaves one computer, passes through up to 30 routers, and then reaches the destination. Return traffic does the same. –At each router, it must decide where to send the packet next. Each router tries to get the packet 1 step closer to the destination.

Which way do we go… Traceroute –Utility which shows you the path between you and a remote computer –Windows: tracert, Unix/Mac: traceroute –Tracert –C:\Documents and Settings\esulliva>tracert Tracing route to [ ] over a maximum of 30 hops: 1 3 ms 3 ms 3 ms fw in-addr.arpa [ ] 2 13 ms 13 ms 12 ms ms 13 ms 14 ms ms 29 ms 28 ms ms 15 ms 16 ms ms 28 ms 28 ms ms 28 ms 27 ms ms 28 ms 28 ms ge p150.msr2.re1.yahoo.com [ ] 9 28 ms 27 ms 28 ms ge-1-41.bas-a1.re3.yahoo.com [ ] ms 27 ms 33 ms f1. [ ] Trace complete.

How do they decide Routers can either be manually programmed –Someone manually enters a routing table which tells it where each packet should go This is useful when you control the network yourself and know where things should go Unmanageable for a very large network Routers can “learn” from neighbors where to route –There are many protocols for doing this: Local Network: RIP, OSPF Internet: BGP

What goes into the decision Assuming a router has more than 1 path to a destination –Many thing may be taken into account in selecting the “best” path Amount of time it takes to get to destination via each path Cost of going over each path Reliability of each path Queue space of each path Evenness of traffic across multiple paths –This is counter intuitive, because you generally want your traffic to get there quickly, but this actually will make it go slower.

Dijkstra’s Algorithm Figure out the shortest path between 2 location by visiting all neighbors of each node: –E.g. To get from A to F: A -> C = 4 C -> F = 3 Total cost A -> F: 7

Dynamic Routing 2 Protocols –Routing Internet Protocol (RIP) Each router talks to every router around it and shares its route table. –Open Shortest Path First (OSPF) Dijkstra’s Algorithm Groups of routers form area Routers flood their information to area Information include speed to next hops

Congestion Various methods supported to fix –Ignore packets once your buffer fills –Send signal to source telling it to slow down –Send signal to destination telling it to tell source to slow down