Introduction to Communication Networks Spring Unit 13 Network extensions Bridges.

Transcription

1 Introduction to Communication Networks Spring 2007 Unit 13 Network extensions Bridges.

2 Acknowledgements slides coming from: The book by Peterson/Davie The book by Wiliam Stallings Slides from lectures by Peter Steenkiste (CMU), also Nick McKeown (Stanford), and Peterson (Princeton) - following the Peterson/Davie book. Slides from the earlier issues of the EECS 122 taught by Prof Jean Walrand. Some slides form the book Andrew Tannenbaum, Computer Networks, 4th edition, the books by Fred Halsall, Farouzan 2of 54

3 Ethernet: Minimum Frame Size reasoning Ethernet you remember the basic assumption? (EPFL) : 64 bytes for 10Mbits/s 3of 54

4 Physical Propagation Limits In the 10 Mbits/s case, the max end-to-end delay is not achievable on a single piece of cable... The limits because of signal shape.. So we use repeaters... The total delay (propagation + delay on all repeaters) on the longest path must NOT exceed the limiting value... Higher bit-rate must be compensated by Either shorter distances (100 Mbits/s ethernet), or Longer data packets (1 Gbit/s Ethernet) In well engineered system errors NOT resulting from collisions are very rare... This does NOT hold in case of complex topologies: numerous bus- segments interconnected via repeaters The star topology is the winner!! 4of 54

5 Hub vs. Switch Collision Domain, Walrand Hub Hub 5of 54

6 Hub vs.switch A (Stallings) Be aware: - The parallelism in transmission holds only with unicast traffic! - Broadcast goes over ALL ports! - Multicast goes over all ports, each receiver makes the decision for himself! What remains out of the ETHERNET MAC on the point-to point link? 6of 54

7 What Remains from the Ethernet MAC... [Farouzan] 7of 54

8 Switched Approach vs. Shared Medium On a shared medium the sender is aware of losses (by detecting collisions) and starts retransmissions For a shared medium this means no losses (up to maximum retransmission number!) - but variable delays (because of retransmissions) In a switched topology, there is a need for buffering and forwarding thus there can be losses (buffer overflow etc.) without detection by the original sender of the packet Higher level protocols have to cope with this problem!! And- finally this does not scale arbitrarily!!! /broadcasts!) See later (e.g. ARP. We do use this..) 8of 54

9 Repeaters, Hubs, Bridges, Switches, Routers and Gateways Tannenbaum 9of 54

10 Bridges, routers...differences on technologies 10 of 54

11 Bridging of LAN s 11 of 54

12 Remote Bridges [Tannenbaum] Remote bridges can be used to interconnect distant LANs. 12 of 54

13 Bridges functions (homogeneous LANs assumed!!!) A configuration with four LANs and two bridges. - Receiving all packets broadcasted within each LAN - The bridge is crossed only if the target station is behind HOW DOES THE BRIDGE KNOW??? - The content of the packet is forwarded without any modification!!!! - only the address is considered! 13 of 54

14 Forwarding of relevant packets 14 of 54

15 Architecture of a bridge [Halsall] 15 of 54

16 Features of Bridges Decrease of the traffic on individual LAN segements in comparison with broadcasting in case of repeater usage. (+) Removal of any physical constraints: both the total number of attached stations and the number of segments making up a LAN can be readily increased. (+) Bridges perform their relaying function based solely on the MAC sub-address in a frame with the effect they are transparent to the protocols being used at higher layers in the protocol stack. This means they can be used with LAN s supporting different protocol stacks. (+) Additional store-and-forward delay in comparison with repeaters. (-) Bridges may overload during periods of high traffic; losses might occur. (-) 16 of 54

17 Bridges how to? Transparent Bridges!! The main idea: the bridge is completely invisible to the stations No changes in the stations No configuration needed in the bridge itself just plug in. Bridges have to learn the topology...by building up the Forwarding Data Base (FDB). An Entry for each host who is origin of any received packet contains the Sender ID; port on which packet has been received and time of receive. Operation after receiving of each packet: If sender already in the FDB update the receive time If sender not yet in the FDB.. Include! The topology of the network might change. Thus entries older that a given time period are removed. THIS IS SO CALLED SOFT STATE CONCEPT!!! 17 of 54

18 The forwarding data base... [steenkinste] 18 of 54

19 What if the address is NOT in the FDB? The assumption- most of the stations WILL send something thus generating an entry in the FDB This is especially true if INTERNET Protocols are used see DHCP next week!!! But if no activity for some period no entry? Packets for unknown stations are just broadcasted on all ports besides of the port of arrival. Let us consider it in a simple configuration 19 of 54

20 Routing with FDB Bridge learning and FDB updating: Station 1 sends a packet P to station 2, but the ID of station 2 does not appear the in the FDB of any active port of bridge 1. Packet P is broadcasted into LAN s B and C by bridge 1, and then it is broadcasted into LAN s D and E by bridge 2. In the process the ID of the node 1 is entered in the left-side FDB s of bridges 1 and 2. Station 2 replies with a packet R, which goes directly to its destination station 1 through bridges 2 and 1, because the ID number of station 1 appears in the left-side FDB s of these bridges. 20 of 54

21 A model of the network Let us consider a configuration with LANS and Bridges... B3 B5 B8 B2 B7 B1 B6 B4 Let us represent this configuration as a graph: bridge= node 21 of 54

22 Flooding Packet broadcasting from node A to all other nodes by (a) pure flooding and (b) a spanning tree. Arrows indicate packet transmissions at the times shown (each packet transmission time is assumed to be one unit long). In general, flooding is an approach to achieve quickest possible distribution of information, independent of the topology. It is also nice in the case of unreliable links... In general more packets are send than really needed of 54

23 Flooding repetitions of packets... Another example: 23 of 54

24 Constrains for flooding Simple flooding implies multiple arrival of a packet... Flooding should not go on infinitely...some limitation is needed Different approaches to constrain the flooding: Lifetime of a packet can be limited (say to N traversed nodes) Nodes could store some parameters of the packet, and deny transmissions which are obviously repetitions. Candidate parameters: Source address, and packet id. Problem: In each case looking into the packet is necessary!!! 24 of 54

25 Addendum: Broadcast Tree Construction Flooding is used to construct the tree rooted at A (shortest path). Each node sends the packet to all its neighbors (besides the origin) All nodes should mark the transmitter of the first packet they receive as their parent on the tree. Nodes should relay the packet to their neighbors only once all subsequent receptions of the same packet should be ignored Now the node know where to send the packet addressed to the root. How should the root reach the node? Source routing...or tables of 54

26 Flooding because of missing FDB entries... Loops! Simple flooding implies multiple arrival of a packet... Flooding should not go on infinitely...some limitation is needed Different approaches to constrain the flooding: Lifetime of a packet can be limited (say to N traversed nodes) Nodes could store some parameters of the packet, and deny transmissions which are obviously repetitions. Candidate parameters: Source address, and packet id. Problem: In each case looking into the packet is necessary!!! 26 of 54

27 Loop of Bridges: B is in the FDB... Both bridges will believe that Station A is on LAN Y!!! 27 of 54

28 Loop of Bridges: B unknown (inexistent?) As none of the bridges looks INTO the packet, they do not know that there are repetitions.. So they will take packets from LAN Y and forward to LAN X of 54

29 Preventing loops Spanning Tree Protocol The network of bridges is a graph. The Spanning Tree Protocol finds a a subgraph that spans all the vertices without loops. Spanning => all bridges are included. Tree => the topology has no loops. The distributed protocol runs: To determine which bridge is the root of the tree, and Each bridge turns off ports that are not part of the tree. 29 of 54

30 Example Spanning Tree [McKeown],modified Each port on a bridge has a unique MAC address. Each bridge has a unique ID e.g. Lowest of the port addresses. B2 B3 B5 B1 B7 B8 Protocol operation: 1. Picks a root lowest ID 2. For each LAN, picks a designated bridge which is closest to the root id used as tie break 3. All bridges on a LAN send packets towards the root via the designated bridge. B6 B4 30 of 54

31 Example Spanning Tree [McKeown] B8 B3 Spanning Tree: B5 B1 B2 B7 B2 B4 B5 B7 B1 Root B8 B6 B4 31 of 54

32 Algorithm Details [Peterson, Princeton] Bridges exchange configuration messages id for bridge sending the message id for what the sending bridge believes to be root bridge distance (hops) from sending bridge to root bridge Each bridge records current best configuration message for each port Initially, each bridge believes it is the root E.g. B5 starts with (B5,B5,0) 32 of 54

33 Algorithm Detail (cont) [Peterson, Princeton] When learn not root, stop generating config messages in steady state, only root generates configuration messages When learn not designated bridge, stop forwarding config messages in steady state, only designated bridges forward config messages Root continues to periodically send config messages If any bridge does not receive config message after a period of time, it starts generating config messages claiming to be the root 33 of 54

34 Spanning Tree Algorithm example [Steenkiste] 34 of 54

35 Bridges from 802.x to 802.y??? [Tannenbaum] Operation of a LAN bridge from to of 54

36 Bridges from 802.x to 802.y??? [Tannenbaum] The IEEE 802 frame formats. The drawing is not to scale. Note the different packet lengths and semantics: (acknowledged transmission versus pure datagram) This is NOT possible without - Putting limits on the packet length (using the shortest ones) - Manipulating the packet It is, however, done in quite a few products. 36 of 54

37 More Problems in Bridging 802.x to 802.y [Tannenbaum] Comment AW: You do not Have to know all these standards just be aware of some other problems!!! 37 of 54

38 WLANs. Extending the coverage/supporting mobility An example how heterogeneity can be hidden Take WLANs with interconnected access points (AP) tethered each mobile node associates with an AP Distribution system AP-1 AP-3 AP-2 F A B G H C C D E 38 of 54

39 Frame format Types control frames, management frames, data frames Sequence numbers important against duplicated frames due to lost ACKs Addresses receiver, transmitter (physical), BSS identifier, sender (logical) Miscellaneous sending time, checksum, frame control, data bytes Duration/ ID Address 1 Address 2 Address 3 Sequence Control Frame Control Address Data 4 bits Protocol version Type Subtype To DS From DS More Frag Power Retry Mgmt More Data WEP Order CRC 39 of 54

40 MAC address format scenario to DS from address 1 address 2 address 3 address 4 DS ad-hoc network 0 0 DA SA BSSID - infrastructure 0 1 DA BSSID SA - network, from AP infrastructure 1 0 BSSID SA DA - network, to AP infrastructure network, within DS 1 1 RA TA DA SA DS: Distribution System AP: Access Point DA: Destination Address SA: Source Address BSSID: Basic Service Set Identifier RA: Receiver Address TA: Transmitter Address 40 of 54

41 How is the distribution system used? It depends which technology is used for the distribution system... If the distribution system is Ethernet, one might add the bridge functionality to APs and break a bit the clean definition of the bridging: Keep the WLAN packets short enough to comply with Ethernet limits Transform in the AP (with bridge functionality) the packet in the Ethernet packet using the Source and Destination addresses only, let other bridges find out the path (as described above) Compute new checksum. Post it to the other AP (if the destination is in the FDB) or broadcast.. The other AP will reformat the Ethernet packet in an packet Not quite clean from the architecture point of view but works! A clean architecture necessary if the differences bigger than in the ETHERNET case see next slide of 54

42 Bridge Protocol Architecture for the remote case of 54

43 Summary: Scalability of Bridges [Steenkiste] 43 of 54

44 Layer 2 Switch v Bridge [Stallings] Layer 2 switch can incorporate logic to function as multi-port bridge (one port of a switch can support a bench of stations!) Bridge only analyzes and forwards one frame at a time Switch has multiple parallel data paths Can handle multiple frames at a time Bridge uses store-and-forward operation Switch can have cut-through operation Bridge suffered commercially New installations typically include layer 2 switches with bridge functionality rather than bridges 44 of 54

45 Motivation for switch with bridging functionality of 54

46 Problems with Layer 2 devices As number of devices in building grows, layer 2 devices (bridges, Switches with bridge functionality) reveal some inadequacies Set of devices and LANs connected by layer 2 switches have flat address space All users share common MAC broadcast address If any device issues broadcast frame, that frame is delivered to all devices attached to network connected by layer 2 switches and/or bridges In large network, broadcast frames can create big overhead Malfunctioning device can create broadcast storm Supporting of the spanning tree topology complex; supporting of multiple simultaneous paths not possible! go for hierarchical address space and more complex routes 46 of 54