TCP/IP Bridging, Switching and Routing in LANs Alvin Kwan

TCP/IP Bridge

TCP/IP Segmenting with Bridges  A bridge is the middle device between network segments within a LAN It aims to reduce data collisions by separating collision domains between segments It aims to reduce data collisions by separating collision domains between segments It works on the data link layer with the use of the MAC address It works on the data link layer with the use of the MAC address It has a buffer (memory) to keep MAC addresses of the network devices in each network segments through a self-configuring process It has a buffer (memory) to keep MAC addresses of the network devices in each network segments through a self-configuring process

TCP/IP How does Bridging work?  A bridge builds an address table that consists of an up-to-date listing of every MAC address on the LAN, as well as the physical bridge port connected to the segment containing that address by listening to all LAN traffic.  When a frame is addressed to an unknown MAC address, the bridge will propagate that frame to all its attached LAN segments (except the segment from which the frame is received).

TCP/IP Important Remarks  A bridge must adhere to the media access protocol, e.g. an Ethernet bridge must conform to the CSMA/CD media access protocol.  No matter how many bridges are in a network, the entire network will share the same logical broadcast address space.  Bridging may increase a latency 10%-30%.

TCP/IP Bridge Types  Some major bridge types are as follows: Transparent bridges Transparent bridges  Link together segments of the same LAN type Speed-buffering bridges Speed-buffering bridges  Similar to transparent bridges except that linked LAN segments may be of different speeds Translating bridges Translating bridges  Similar to transparent bridges except that linked LAN segments may be of different LAN types

TCP/IP Bridging Today  Most bridges are two-port device though multi- port bridges are available too.  Bridges are becoming obsolete because their functions have been almost “subsumed” by other networking devices, noticeably LAN switches.  Hubs with high performance up-link ports are in fact bridges in disguise.

TCP/IP Exercise  Connecting network devices arbitrarily to different network segments separated by a bridge will not be able to make the best use of a bridge. Why?

TCP/IP Switch

TCP/IP Segmenting with Switches  A switch is the multiport data link layer device It aims to reduce data collisions by associating each port with its own collision domain It aims to reduce data collisions by associating each port with its own collision domain It uses the MAC address It uses the MAC address It dynamically builds and maintains a MAC filtering table, holding all of the necessary MAC information for each port It dynamically builds and maintains a MAC filtering table, holding all of the necessary MAC information for each port

TCP/IP Switching Issues  Address learning (when a switch is just turned on OR change in network configuration)  Forward/filter decision  Loop avoidance

TCP/IP Address Learning (1/3)

TCP/IP Address Learning (2/3)

TCP/IP Address Learning (3/3)

TCP/IP Forward/Filtering Decisions  When a frame arrives at a switch, the switch checks the destination hardware address, which is compared to the forward/filter MAC database. If the destination hardware address is known, then it will transmit it out the correct port, but if the destination hardware address is not known, then it will broadcast the frame out of all ports, except the one which it received it from. If a device (computer) answers to the broadcast, then the MAC address of that device is added to the MAC database of the switch.

TCP/IP Loop Avoidance  Any potential problem with the following network configuration?

TCP/IP Store & Forward Mode  When the switch receives a frame from one of it's ports, it will store it in memory, check it for errors and corruption, and if it passes the test, it will forward the frame out the designated port, otherwise, if it discovers that the frame has errors or is corrupt, it will discard it.  This method is the safest, but also has the highest latency.

TCP/IP Cut-through Mode  The switch reads the frame until it learns the destination MAC address of the frame it's receiving. Once it learns it, it will forward the frame straight out the designated port without doing any error checking.

TCP/IP Memory Requirement for Switch

TCP/IP Routing

TCP/IP Routing  Concerns with “learning how to get from here to there”  Works on the network layer, i.e. Layer 3.  Two major camps Source routing (less common) Source routing (less common) Hop-to-hop routing (predominately used in TCP/IP networks) Hop-to-hop routing (predominately used in TCP/IP networks)

TCP/IP Routing Principles  Correct route  Most direct route Shortest route Shortest route Route takes the least time Route takes the least time  Most reliable route (which may not be the shortest one)

TCP/IP Routing Requirements over Internet  Static routing cannot help; good routing has to be dynamic dynamic adaptive adaptive decentralized decentralized scale well, and scale well, and resilent resilent

TCP/IP Source Routing  Routing information is collected by the source  Routing information is put into the packets that the source launches toward the destination  Intervening network (with intermediate links and systems) read the routing information from the packets and act on it accordingly  Example: route planning by most people is a kind of source routing

TCP/IP Hop-by-hop Routing  Hop-by-hop routing requires routing protocols that allow end systems and intermediate systems to collect and distribute the information necessary to determine routes routing protocols that allow end systems and intermediate systems to collect and distribute the information necessary to determine routes a routing information base containing information from which routes between end systems can be computed, and a routing information base containing information from which routes between end systems can be computed, and a routing algorithm that uses the information contained in the routing information base to derive routes between end systems a routing algorithm that uses the information contained in the routing information base to derive routes between end systems

TCP/IP Example: Hop-by-hop Routing

TCP/IP Example: Simple Routing  Connected network

TCP/IP Example: Simple Routing  Connected network (no routing is required)  Q: What is the range of addresses that Node B can take such that it can “talk” to Node A?

TCP/IP Example: Simple Routing  Unconnected network As A and C are separately connected to different networks, they are unconnected. As A and C are separately connected to different networks, they are unconnected. Without setting up routing, no communications can be achieved between A and C. Without setting up routing, no communications can be achieved between A and C.

TCP/IP Example: Routing Table

TCP/IP Example: Routing Table

TCP/IP Example: Routing Table

TCP/IP Example: Data Link Frame

TCP/IP

Exercise  Suppose we would like to improve the effective bandwidth of a LAN with no subnet definition, would a switch be more favorable than a router? Why?

TCP/IP Fault Tolerance of Switch

TCP/IP Fault Tolerance of Router (1/2)

TCP/IP Fault Tolerance of Router (2/2)

TCP/IP Broadcast Flood (1/2)

TCP/IP Broadcast Flood (2/2)

TCP/IP Important Remarks  Switching builds logically fat networks whereas routing builds logically hierarchical networks  Switches segment LANs whereas routers tend to segment WANs

TCP/IP Exercise (source: inf.mpg.de/~weidenb/Lan2004/exercise8.pdf)

TCP/IP References  Wikipedia’s pages on network switch and router  /presentations/routing-smith.pdf /presentations/routing-smith.pdf /presentations/routing-smith.pdf  asia/pkg1/06/index_bar.html asia/pkg1/06/index_bar.html asia/pkg1/06/index_bar.html  html html html