Packet Switching Around 1970 research began on a

Packet Switching Around 1970, research began on a new form of architecture for long distance communications: Packet Switching. 1

Introduction z. Packet Switching refers to protocols in which messages are divided into packets before they are sent. Each packet is then transmitted individually and can even follow different routes to its destination. z. Once all the packets forming a message arrive at the destination, they are recompiled into the original message. 2

Packet Switching Application z Most modern Wide Area Network (WAN) protocols, including TCP/IP, X. 25, and Frame Relay, are based on packet-switching technologies. In contrast, normal telephone service is based on a circuit-switching technology, in which a dedicated line is allocated for transmission between two parties. z Circuit-switching is ideal when data must be transmitted quickly and must arrive in the same order in which it's sent. This is the case with most real-time data, such as live audio and video. z Packet switching is more efficient and robust for data that can withstand some delays in transmission, such as e-mail messages and Web pages. 3

Recall Circuit Switching z. Call Set-up z. Data Transfer z. Call disconnect 4

Features of Circuit Switching z. Circuit switching is connection oriented. z. Resources are allocated for the call throughout the network. z. Calls may be blocked if the resources are not available. z. Circuit Switching originated due to need for voice communications. 5

Circuit Switching for Data z When Circuit Switching networks started to be used for data communications it became clear that: y. In circuit switching resources dedicated to a particular call whereas data traffic is bursty so most of the time allocated resources would be unutilized. y. Data rate is fixed in circuit switching so both ends must operate at the same rate - whereas there is asymmetry in the data rate required for each communicating party for data communication needs. 6

Packet Switching z. Packet Switching started in the 1970 s. z. ARPANET that became Internet z. In the beginning most people did not believe it would work z. The basic technology of packet switching is fundamentally the same today as it was in the early 1970’s networks z. Packet switching remains one of the few technologies for effective long-distance data communications. 7

Packet Switching Operation z Data are transmitted in short packets. Typically an upper bound on packet size is 1000 octets. z If a station has a longer message to send it breaks it up into a series of small packets. Each packet now contains part of the user's data and some control information. z The control information should at least contain: y. Destination Address y. Source Address z Store and forward - Packets are received, stored briefly (buffered) and past on to the next node 8

Packet Switching Operation 9

Use of Packets 10

Packet Switching Networks are Switched Networks 11

Advantages z. Line efficiency y. Single node to node link can be dynamically shared by many packets over time y. Packets queued and transmitted as fast as possible z. Data rate conversion y Two stations of different data rates can exchange packets because each connects to its node at its proper data rate 12

Advantages z. When traffic becomes heavy in a circuit switching network, some calls are blocked i. e the network refuses to accept additional connection requests until the load on the network decreases. z. On a packet switching network, packets are still accepted, but delivery delay increases z. Priorities can be used. If a node has a number of packets queued for transmission, it can transmit the higher priority packets first. 13

Switching Technique z. If a station sends a message through packet switching network that is of length greater than the maximum packet size, it breaks the message up into packets and sends these packets, one at a time, to the network z. Question? ? How the network will handle this stream of packets as it attempts to route them through the network and deliver them to the intended destination 14

Switching Technique - Virtual Circuits and Datagrams z. Packets handled in two ways y. Datagram approach y. Virtual circuit approach 15

Datagram Packet Switching z In datagram approach each packet is treated independently with no reference to packets that have gone before. No connection is set up. z Packets can take any practical route z Packets may arrive out of order z Packets may go missing z Up to receiver to re-order packets and recover from missing packets z More processing time per packet per node z Robust in the face of link or node failures. 16

Packet Switching Datagram Approach 17

Virtual Circuit Packet Switching z. In the Virtual Circuit approach a pre-planned route is established before any packets are sent. z. There is a call set up before the exchange of data (handshake). z. All packets follow the same route and therefore arrive in sequence. z. Each packet contains a virtual circuit identifier instead of destination address z. More set up time 18

Virtual Circuit Packet Switching z. No routing decisions required for each packet - Less routing or processing time z. Susceptible to data loss in the face of link or node failure z. Clear request to drop circuit z. Not a dedicated path 19

Packet Switching Virtual Circuit Approach 20

Virtual Circuit 21

One Station Can Have Many Virtual Circuit Connections 22

Virtual Circuits vs. Datagram z. So the main characteristic of the virtual circuit technique is that a route between stations is setup prior to data transfer, this does not mean that this is a dedicated path as in the circuit switching z. A packet is still buffered at each node, and queued for output over a line, while other packets on other virtual circuits may share the use of the line 23

Virtual Circuits vs. Datagram z. Virtual circuits y. Network can provide sequencing and error control y. Packets are forwarded more quickly x. No routing decisions to make y. Less reliable x. Loss of a node looses all circuits through that node z. Datagram y. No call setup phase x. Better if few packets y. More flexible x. Routing can be used to avoid congested parts of the network 24

Packet Size 25

Packet Size z In this example it is assumed that there is a virtual circuit from station X through nodes a and b to station Y. z The message to be sent comprises 40 octets and 3 octets of control information called header. z If the entire message is sent the packet first transmitted from station X to node a, when the entire packet is received, it can be transmitted from a to b and then transmitted to Y. ignoring switching time, total transmission time is 129 octettime(43 octets x 3 packet transmission) 26

Circuit vs. Packet Switching z. Performance y. Propagation delay x. The time it takes a signal to propagate from one node to the next. This time generally negligible. Typically on a wire medium 2 x 108 y. Transmission time x. The time it takes for a transmitter to send out a block of data, e. g it takes 1 s to transmit 10, 000 bit block of data onto a 10 -kbps line. y. Node delay x. The time it takes for a node to perform the necessary processing as it switches data 27

Comparison with Circuit Switching - Event Timing 28

Circuit Switching Packet Switching Datagram Packet switching Virtual-circuit Packet Switching Dedicated transmission path No dedicated path Continuous transmission of data Transmission of packets Fast enough for interactive Messages are not stored Packets may be stored until delivered Packets stored until delivered The path is established for entire conversation Route established for each packet Route established for entire conversation Call setup delay; negligible transmission delay Packet transmission delay Call setup delay; packet transmission delay Busy signal if called party busy Sender may be notified if packet not delivered Sender notified of connection denial Overload may block call setup; no delay for established calls Overload increases packet delay Overload may block call setup; increase packet delay Electromechanical or computerized switching Small switching nodes User responsible for message loss protection Network may be responsible for individual packets Network may be responsible for packet sequences Usually no speed or code conversion Speed and code conversion Fixed bandwidth Dynamic use of bandwidth No overhead bits after call setup Overhead bits in each packet 29

Packet switching datagrams or virtual circuits z. Interface between station and network node y. Connection oriented x. Station requests logical connection (virtual circuit) x. All packets identified as belonging to that connection & sequentially numbered x. Network delivers packets in sequence x. External virtual circuit service xe. g. X. 25 x. Different from internal virtual circuit operation y. Connectionless x. Packets handled independently x. External datagram service x. Different from internal datagram operation 30

Combinations (1) z External virtual circuit, internal virtual circuit y. When the user requests a virtual circuit, a dedicated route through the network is constructed. y All packets follow the same route z External virtual circuit, internal datagram y. Network handles each packet separately y. Different packets for the same external virtual circuit may take different internal routes y. Network buffers at destination node for re-ordering 31

Combinations (2) z External datagram, internal datagram y. Each packets is treated independently from both the user’s and the network’s point of view z External datagram, internal virtual circuit y. External user does not see any connections simply sending packets one at a time y. Network sets up logical connection between stations for packet delivery 32

External Virtual Circuit and Datagram Operation 33

Internal Virtual Circuit and Datagram Operation 34

External and Internal Operation - ED/IVC 35

External and Internal Operation - ED/ID 36

External and Internal Operation - EVC/IVC 37

External and Internal Operation - EVC/ID 38

Packet switching - datagram v/s virtual circuits z. The datagram service, coupled with internal datagram operation, allows for efficient use of the network z. There is no call setup and no need to hold up packets while a packet in error is retransmitted z. The virtual circuit service can provide end-toend sequencing and error control 39

Packet switching - datagram v/s virtual circuits z virtual circuit is attractive for supporting connection-oriented applications such as file transfer and remote terminal access z. In practice, the virtual circuit service is much more common than the datagram service z. The reliability and convenience of a connection-oriented service is seen as more attractive than the benefits of the datagram service 40

Routing z. Complex, crucial aspect of packet switched networks z. Characteristics required y. Correctness y. Simplicity y. Robustness y. Stability y. Fairness y. Optimality y. Efficiency 41

Routing Performance Criteria z. The selection of a route is generally based on some performance criterion. z. Minimum hop z. Least cost Routing y. Cost is associated with each link y. Cost could be inversely related to the data rate y. Using some algorithm z. Delay z. Throughput 42

Routing Decision Time and Place z Routing decisions are made on the basis of some performance criterion. Two key characteristics of the decision are the time and place that the decision is made. z Decision time is determined by whether the routing decision is made on a packet or virtual circuit basis. z When the internal operation of the network is datagram, a routing decision is made individually for each packet. z For the internal virtual circuit operation, a routing decision is made at the time the virtual circuit is established 43

Routing Decision Time and Place z. The term decision place refers to which node or nodes in the network are responsible for the routing decision. y. Distributed x. Made by each node (more complex but more robust) y. Centralized (network control centre) x. Loss of the control centre may block operation of the network y. Source routing x(decision is actually made by the source station rather than by a network node) 44

Network Information Source z Routing decisions usually based on knowledge of network (not always) z Some strategies use no such information and yet manage to get packets through; flooding and some random strategies z Distributed routing y. Nodes use local knowledge y. May collect info from adjacent nodes y. May collect info from all nodes on a potential route z Central routing y. Central node collect info from all nodes 45

Network Information Update Timing z. A related concept is that of information update timing y. Which is a function of both the information source and the routing strategy y. If no information is used (as in flooding), there is no information to update y. Fixed strategy – information is never updated y. Adaptive strategy – information is updated from time to enable the routing decision to adapt to changing decision (regular updates) 46

Elements of Routing Techniques for Packet-Switching Networks z Performance Criteria y Number of Hops y Cost y Delay y Throughput z Decision time y Packet (Datagram) y Session (Virtual Circuit) z Decision Place y Each node (distributed) y Central node (Centralized) y Originating node (Source) z Network Information Source y None y Local y Adjacent node y Node along route y All nodes z Network information update y Continuous y Periodic y Major load change y Topology change 47

Routing Strategies z. Fixed z. Flooding z. Random z. Adaptive 48

Fixed Routing z. Single permanent route for each source to destination pair z. Determine routes using a least cost algorithm (appendix 10 A) z. Route fixed, at least until a change in network topology 49

Fixed Routing Tables 50

Fixed Routing z. With fixed routing, there is no difference between routing for datagrams and virtual circuits z. All packets from a given source to a given destination follow the same route z. The advantage of fixed routing is simplicity and it should work well in a reliable network with a stable load z. Its disadvantage is its lack of flexibility. It does not react to network congestion or failures 51

Flooding z No network info required z Packet sent by node to every neighbor z Incoming packets retransmitted on every link except incoming link z Eventually a number of copies will arrive at destination z Each packet is uniquely numbered so duplicates can be discarded z Nodes can remember packets already forwarded to keep network load in bounds z Can include a hop count in packets 52

Flooding Example 53

Properties of Flooding z. All possible routes between source and destination are tried y. Very robust (used to send emergency messages) z. Because all routes are tried at least one copy of the packet to arrive at the destination will have used a minimum hop route z. All nodes are visited y. Useful to distribute information (e. g. routing) z. The principal disadvantage of flooding is the high traffic load 54

Random Routing z Random routing has the simplicity and robustness of flooding with far less traffic load z Node selects one outgoing path for retransmission of incoming packet z Selection can be random or round robin z Can select outgoing path based on probability calculation z No network info needed z Route is typically not least cost nor minimum hop 55

Adaptive Routing z Used by almost all packet switching networks z Routing decisions change as conditions on the network change y. Failure x. When a node or trunk fails, it can no longer be used as part of a route y. Congestion x. Portion of a network is heavily congested, packets route around rather than the area of congestion z Requires info about network z Decisions more complex (processing burden on network nodes increases) 56

Adaptive Routing - Advantages z. Reacting too quickly, causing congestionproducing oscillation, or too slowly, being irrelevant z. Improved performance, as seen by the network user z. An adaptive routing strategy can aid in congestion control 57

ARPANET Routing Strategies(1) z First Generation y 1969 y. Distributed adaptive y. Estimated delay as performance criterion y. Bellman-Ford algorithm y. Node exchanges delay vector with neighbors y. Update routing table based on incoming info y. Doesn't consider line speed, just queue length y. Queue length not a good measurement of delay y. Responds slowly to congestion 58

ARPANET Routing Strategies(2) z. Second Generation y 1979 y. Uses delay as performance criterion y. Delay measured directly y. Uses Dijkstra’s algorithm y. Good under light and medium loads y. Under heavy loads, little correlation between reported delays and those experienced 59

ARPANET Routing Strategies(3) z. Third Generation y 1987 y. Link cost calculations changed y. Measure average delay over last 10 seconds y. Normalize based on current value and previous results 60

Costing of Routes 61

Dijkstra’s Algorithm z. Define: z N = set of all nodes in the network z s = source node z M = set of nodes so far incorporated by the algorithm z dij = link cost from node i to node j; dii = 0 and dij = if the two nodes are not directly connected; dij 0 if the two nodes are directly connected z Dn = cost of the least cost path from node s to node n that is currently known to the algorithm 62

63

Example Algorithm Dijkstra 64

Results for Dijkstra Algorithm It er ati on T L(2) Path L(3) Path L(4) Path L(5) Path L(6 ) Path 1 {1} 2 1– 2 5 1 -3 1 1– 4 - - 2 {1, 4} 2 1– 2 4 1 -4 -3 1 1– 4 2 1 -4– 5 - 3 {1, 2, 4} 2 1– 2 4 1 -4 -3 1 1– 4 2 1 -4– 5 - 4 {1, 2, 4, 5} 2 1– 2 3 1 -4 -5– 3 1 1– 4 2 1 -4– 5 4 1 -4 -5– 6 5 {1, 2, 3, 4, 5} 2 1– 2 3 1 -4 -5– 3 1 1– 4 2 1 -4– 5 4 1 -4 -5– 6 65

Bellman-Ford Algorithm Definitions z Find shortest paths from given node subject to constraint that paths contain at most one link z Find the shortest paths with a constraint of paths of at most two links z And so on z s = source node z w(i, j) = link cost from node i to node j yw(i, i) = 0 yw(i, j) = if the two nodes are not directly connected yw(i, j) 0 if the two nodes are directly connected z h = maximum number of links in path at current stage of the algorithm z Lh(n) = cost of least-cost path from s to n under constraint of 66 no more than h links

Example of Bellman-Ford Algorithm 67

Results of Bellman-Ford Example h Lh(2 Pat Lh(3 Path ) Lh(4 Pat Lh(5 Path Lh(6 Path ) ) 0 - - - 1 2 1 -2 5 1 -3 1 1 -4 - - 2 2 1 -2 4 1 -4 -3 1 1 -4 2 1 -4 - 10 5 1 -3 -6 3 2 1 -2 3 1 -4 -5 - 1 3 1 -4 2 1 -4 - 4 5 1 -4 -5 -6 4 2 1 -2 3 1 -4 -5 - 1 3 1 -4 2 1 -4 - 4 5 1 -4 -5 -6 68

Comparison z Results from two algorithms agree z Information gathered y. Bellman-Ford x. Calculation for node n involves knowledge of link cost to all neighboring nodes plus total cost to each neighbor from s x. Each node can maintain set of costs and paths for every other node x. Can exchange information with direct neighbors x. Can update costs and paths based on information from neighbors and knowledge of link costs y. Dijkstra x. Each node needs complete topology x. Must know link costs of all links in network x. Must exchange information with all other nodes 69

Evaluation z Dependent on processing time of the algorithms and the amount of information that must be collected from other nodes in the network z The evaluation will be depend on the implementation approach and the specific implementation z Both algorithms are known to converge under static conditions of topology and link costs and will converge to the same solution z If the link costs change over time, the algorithms will attempt to catch up with these changes z However, if the link costs depend on traffic, which in turn depends on routes chosen, then a feedback conditions exists, and instabilities may result 70

Packet Switching Evolution z. X. 25 packet-switched network z. Router-based networking z. Switching vs. routing z. Frame relay network z. ATM network 71

Switching v/s Routing z Switching z path set up at connection time z simple table look up z table maintainance via signaling z no out of sequence delivery z lost path may lose connection z much faster than pure routing z link decision made ahead of time, and resources allocated then z Routing z can work as connectionless z complex routing algorithm z table maintainance via protocol z out of sequence delivery likely z robust: no connections lost z significant processing delay z output link decision based on packet header contents - at every node 72