Contents. Bridging. Transparent Bridging (TB) 1. Transparent Bridging Bridging in General. hub

Transcription

1 Contents ÉCOLE POLYTECHNIQUE FÉÉRLE E LUSNNE Bridging Jean-Yves Le Boudec Fall 007!.Transparent bridging!. Spanning Tree Protocol (STP) "#specification $#an exotic version of Bellman-Ford %#the STP protocol, main &#topology changes '#synchronization with forwarding (#efficiency!. Rapid Spanning Tree Protocol (RSTP). Transparent Bridging Bridging in General! Bridges are intermediate systems that forward MC frames to destinations based on MC addresses! Interconnect systems beyond one LN segment, keeping main characteristics of LN without additional addresses MC addresses used to identify end systems preserve sequence integrity! The LN segments can be of different nature Ex: WiFi and Ethernet Transparent Bridging (TB)! End systems ignore that there are transparent bridges bridge is transparent MC frames not changed by bridges frames not sent to bridge, but rather: bridge is promiscuous (listens to all frames)! Bridges are required to be plug and play (i.e. no configuration by system manager)! Q. Is an Ethernet hub a bridge or a repeater? What s the difference? hub! There are several possible methods, only one is wide-spread: Transparent Bridging hub hub 4

2 Transparent Bridging uses forwarding tables Bridges learn addresses by observing traffic! Table maps MC addresses to port numbers No IP addresses here! B C Forwarding Table Repeater port Bridge port port B C Forwarding Table est Port MC Nb addr B C port Bridge port port B C 5! How can a bridge build its table? No equivalent to routing protocols, we need a plug and play! Bridge builds routing table by reading all traffic table built by learning from S field in MC frame learnt addresses times out if not re-learnt! If destination address not in table broadcast to all ports same for group addresses Can this method of learning addresses be extended to a network of bridges? The method of learning does not work if there are loops in the topology! On this example, yes.! Q. How does B see the network?! Q. What happens when send a frame to B? assume empty forwarding tables at the beginning port port port port B port port B B Y C Z port port B port port B B Y C Z 7 8

3 Transparent Bridges force the ctive Topology to be loop-free! Learning bridge works well if there is no loop in the topology The topology can be represented as a bidirectional graph where vertex = bridge, edge = connection through collision domains (called here: LN) for such graphs : Loop- free and connected tree On a tree, there is only one path from one host to one bridge. Therefore, a bridge sees a host on exactly one port.! network of bridges may have redundant connections (as in previous example). This is good for reliability, but this causes loops in the topology.! The adopted by transparent bridges is: maintain an active topology that is loop-free i.e. decide that some ports are blocked This should be done automatically, without configuration (plug and play) The Spanning Tree Protocol! What does it do? Prevent loops in the active topology ecide which ports should be blocked or opened ports that are allowed to forward frames are said to be in the forwarding state or called forwarding ports dapt to changes in the physical topology! How does it work? See next section 9 0 Summary: what a transparent bridge does. The Spanning Tree Protocol Individual PU forwarding Copy all frames on all forwarding ports Frame received on port i -> /* port i is forwarding */ If is unicast, is in forwarding table with port j and j is a forwarding port then copy to port j else flood all forwarding ports! i Update forwarding table with (i, S) We present the Spanning Tree Protocol in 4 steps : )"*Specification )$*esign of main algorithm )%*Main Protocol )&*Topology Changes and Synchronization with Packet Forwarding Control Method Run the Spanning Tree Protocol

4 (a) Specification )"*We now specify the STP method (ie what it does, in more details than before, not how)! There are many ways to build a tree on a graph. Minimum Spanning Tree (Kruskal or Prim s algorithms) The STP chose to use the set of shortest paths towards some selected vertex.! Each bridge has a bridge label, based on MC address + configurable offset. Bridge with smallest label is selected and called root. Each LN between bridges has a cost, by default, decreasing function of bit rate: Port Type uplex Cost 00BSE-T / 00BSE-F (VLT) Full 5 Half 0BSE-T Full Half 700! What: The STP computes a tree of shortest paths to the root bridge Specification of STP (cont d)! STP gives a role to all ports Root or designated (ports on spanning tree) Blocked (ports not on spanning tree)! Root ports One per bridge := port towards root along shortest path in case of equal costs, lowest port id chosen! esignated ports On every LN ( collision domain), choose one designated bridge all ports on LN for which the bridge is designated are designated ports esignated bridge one per LN defined by : it has the shortest path to root possibly root itself! Ports other than root or designated are blocking 4 Understanding the Specification! Q. find the root, root ports, designated bridges, designated ports and blocking ports! Q. find the forwarding table at all bridges T B84 B8 B90 B4 cost = Z (b) STP: design of main algorithm! STP uses a variant of the Bellman-Ford algorithm (see dv.ppt), which we call the Bellman-Ford algorithm for Bridges! Like the original Bellman-Ford algorithm, it is a distributed algorithm, but it is easier to understand it by studying first the centralized version Y B9 B99 cost = 5

5 The Centralized Bellman Ford lgorithm for Bridges uses Special Link and Path ttributes! Represent the network by a graph, where a vertex is a bridge. For this algorithm, we treat the graph as directed.! We are given link costs c(i,j) (costs of LNs, see later for default values) We assume c(i, j) > 0 and c(i,j) = " when i and j are not connected.! We are also given vertex labels l(i)(concatenation of bridge priority and serial number, set by manufaturer)! efinition: a link attribute is (i,j) := [l(j), c(i,j)] concatenation of attributes: [l,c] [l, c ] = [min(l,l ), c+c ] the attribute of a path i i ik is the concatenation of the attributes of the links i.e.: [minimum label, sum of costs] comparison of attributes lexicographic: [l, c]! [l c ] iff [(l < l ) or (l = l and c! c )] a total order on N x [0, "] path p is better than p if attribute of p attribute of p Examples of Path ttributes! Q. Q. What are the attribute of the following paths? Which one is best? 0! 50! ! 0! 0! 40 0! 0! ! 50! 40! What are Best Paths in this Setting?! ssume the graph is fully connected; all vertex labels are different; all link costs are > 0! path p that starts at is best (among all paths that start at i) iff It goes through the vertex i 0 that has the smallest label in the graph (the minimum label is reached at only one vertex, by hypothesis) It stops at i 0 It is a shortest path from i to i 0! Thus: the best paths in this graph are the shortest paths to the node with the smallest label The Centralized Bellman-Ford lgorithm for Bridges! What: Given a directed graph with links attributes as above, computes one tree of best paths from any vertex let (i,j):= attribute of link (i,j) =[l(j), c(i,j)]! How: efine p k (i) as the cost of the best path from i to anywhere in at most k hops.! Theorem +#If the graph is fully connected, the algorithm stops at the latest at k=number of vertices ; at the end, p k (i) is the attribute of a best path,# best path from is obtained by letting pred(i) = the index (j or i) that achieves the minimum in () If the min is achieved by the term [l(i),0] then pred[i]=i; this happens only when vertex i has the smallest label 9 0

6 ! The algorithm is the same as the classical Bellman-Ford algorithm [dv.ppt], with the following modifications Exotic algebra instead of usual algebra: costs are replaced by attributes; addition of costs is replaced by concatenation ( ) and comparison by the lexicographic order. ll paths instead of paths to a specific node: add a virtual node 0 such that (i,0)=[l(i), 0] and (0,i)=[, ]. pply the classical Bellman-Ford to compute the shortest (i.e. best) paths from all nodes i to node 0. Remove the final edge from these paths and obtain the best paths we are looking for. Indeed, with these modifications, the classical Bellman-Ford becomes Run the Centralized Bellman Ford lgorithm for Bridges on this Example! Q. Write p k (i), pred(i) and draw the spanning tree One can easily see that () is equivalent to (), given that we set p k- (0) to [, 0], and that the impact of the initialization for p 0 (i) disappears after one step.! Note: in the algorithm, min is the lexicographic min (derived from the comparison of attributes)! The proof of the algorithm is similar to the classical case. It relies on the fact that is associative. Impact of Initial Conditions on the Bellman-Ford lgorithm for Bridges! The classical Bellman-Ford algorithm continues to work if we take different initial conditions but the interpretation that p k (i) is the distance from i to in at most k hops is no longer true! oes this still hold for the Bellman-Ford algorithm for Bridges? Q. write p k (i), pred(i) and draw the spanning tree, with initial conditions as shown. The dotted link does not exist in the current configuration. It existed before, and explains why node 0 starts with these initial conditions. 09 p k (i): (format: label, cost) k \ i ,0 0,0 09, 40,0 50, The Bellman-Ford lgorithm for Bridges is sensitive to initial conditions Theorem If the initial conditions in the centralized Bellman-Ford lgorithm for Bridges satisfy: 8 i : p 0 (i)=(m i, c i ) with m i # min j l(j) the algorithm converges to the correct value else the algorithm diverges with lim k! p k (i)=(m 0,) where m 0 =min i m i Proof: first show that the label converges to the minimum of all initial conditions (it can only decrease). Then use the property of Bellman-Ford in the usual algebra (see chapter distance vector ) Comment: the convergence may be much longer than with the initial conditions in theorem ll-path variant of Bellman Ford Note that there is a condition on the initial label, not on the initial cost. 4

7 Example! Q. write p k (i), pred(i) and draw the spanning tree, with initial conditions as shown. The dotted link does not exist in the current configuration. It existed before, and explains why node 0 starts with these initial conditions. oes the algorithm converge to the correct values? p k (i): (format: (label, cost)) k \ i ,0 0,0 0, 40,0 50,0 istributed Bellman-Ford lgorithm for Bridges! Like the classical Bellman-Ford (i.e. BF in dv.ppt), the Bellman-Ford lgorithm for Bridges can be distributed: It is the algorithm used by STP every node, say i, maintains an estimate q(i) of p(i), the attribute of a best path from i and of pred(i), the next node on a best path; initially q(i)=[l(i),0] and pred(i)=i from time to time, i sends its value q(i) to all its neighbours when node i receives a value q(j0) from any neighbour j0, it sets q(j0) to the received value and updates q(i) by recomputing eq () if j0 == pred(i) then q(i):=min { (i,j 0 ) q(j 0 ), [l(i),0] } else q(i) := min{(i,j 0 ) q(j 0 ), q(i) } 5 if eq () causes q(i) to be modified, pred(i) is set to j 0 if (i,pred(i)) changes (including if pred(i) stops being a neighbour ) then q(i) is set to [l(i),0] and pred(i) is set to i. Sample Run of the istributed Bellman-Ford lgorithm for Bridges The istributed Bellman-Ford lgorithm for Bridges may need to be reset > > 0 0 -> > > > > > > > 0 link breaks 50 -> 0 0 -> > 0 possible run : i ,0 0,0 0,0 40,0 50,0 0, 0, 0,4 0,7 0,5 0,4 0, 0, 0,4 50 does as if received q(40)= (", "); pred(50)=40 thus 50 does q(50)=(50,0); similarly 40 does a new computation but this does not change 40 0, 50,0 0,0 0,4 0,5! Like the centralized algorithm, the distributed algorithm is robust to changes in configuration as long as the node with the smallest label (called root bridge ) is still present and reachable from all bridges. If this is not true, the algorithm does not converge to a true value. It needs to be reset by some additional mechanism. Q. Compare to the classical distributed Bellman-Ford algorithm. 7 8

8 (c) STP Main Protocol! Standardized by IEEE 80.! ll bridges run it! Implements the istributed Bellman Ford lgorithm for Bridges best attribute = [root bridge label, distance to root bridge] pred(i) is kept in the form of root port instead of bridge label bridge periodically sends its values to neighbours, and whenever a change occurs! We explain by examples in the next slides. Bridge PUs! STP uses control frames called Bridge PUs (BPUs) MC = all bridges (multicast) 0 80 C SP = SP is a field in the IEEE format, which replaces the Ethertype in the Ethernet format.! BPUs are not forwarded by bridges unlike all other frames BPUs are sent by one bridge to all bridges on the same LN segment reminder: a data frame is never sent to bridge by end system! Configuration BPU contains root Id (priority) cost to root (from sender of config BPU) id of sender port number (omitted in the examples) 9 0 Back to Example Spanning Tree Protocol seen by B90! Bridge B90 prepares config BPU and sends on all ports; B90 configuration tables: B8 B4 cost = < , > T B84 B90 Z < > Y B9 B99 cost = Root Port : esignated Ports :, Blocking Ports : Root Port : esignated Ports : Blocking Ports : message received on port : < message format: rootid.cost_to_root.senderid

9 ! Bridges store information per port instead of keeping the info q(i), pred(i). The stored information on every port is the last sent Config BPU, or the last received, whichever happened last.the config BPU format is shown on the figure.! When receiving a message the bridge runs the distributed bellman ford algorithm, updates the states of ports, and decides whether to send a new message on every port:! Initially, the bridge has the same config on all ports. This means that q(90)=[90, 0] and pred(90)=90. We now give explanations to the arrows to 4 on the previous page +#On receiving on port : port is not the root port, in other words, the message is not received from pred(90), thus we are in the else part of Eq (). 4. < 90.0 thus q(90)=[4,] and pred(90)= 4 the bridge stores this information indirectly by storing on port,#since the local state q(90) has changed, an update should be sent to all neighbours, in fact, here the algorithm uses an optimization that exploits the facts that the LNs are broadcast medium. The optimization is that we assume all neighbours that are on port have also received the message we received, so we do not need to forward any new information to this port. the update is 4..90, which means the state information is q(90)=[4,] and it is sent by bridge 90. The information on all ports allows the bridge to know that pred(90)=4 -#On receiving on port : port is not the root port, in other words, the message is not received from pred(90), thus we are in the else part of Eq (). 4. < 4. thus q(90)=[4,] and pred(90)= 99 the bridge stores this information indirectly by storing on port.#since the local state q(90) has changed, an update is sent to all neigbours. Here too, an optimization is done. s before, no update is sent on port, since the last message arrived on this port. Further, the last message received on port was better than the one we would send 4. > 4.0 thus we do not send it on port. The assumption here is that all bridges on port have already received the information ! The figure also shows the status of the ports at instants, marked and the port status is computed by comparing the state information on all ports at : the root port is (this follows from the Bellman-Ford algorithm, as explained above) the bridge with the shortest path to root on ports and is this bridge,s since we sent a config BPU on these two ports. Hence this bridge is designated bridge on the LNs behind ports and, and ports and are designated ports for this bridge. at : the root port is port is designated port is neither root nor designated hence is blocking (d) Topology changes! Topology changes occur due to changes in configuration failures, recoveries! It changes occur, the behaviour depends on whether the root bridge is still reachable from all bridges if so, let distributed Bellman Ford do the job else, we need some additional mechanism: STP uses root monitoring for this: root refreshes validity of STP by periodically sending a refresh message every HelloTime (s) the refresh message is propagated along the spanning tree a bridge that does not receive refresh message for Maxge restarts STP basic procedure from fresh initial conditions (= reset) 5

10 Topology Change That is Handled by Bellman-Ford B99 crashes; we look at B90 B90 detects absence of B99 (absence of hello, or other mechanism); this is equivalent to receiving (in Bellman-Ford s algorithm) a state information: from B90: best attribute (", ") a new local computation is started: timeout on [], > < Topology Change That Requires Reset! If B4 dies, what happens? Root Port : esignated Ports : Blocked Ports : Q. What is the spanning tree after the failure? 7 8 (e) Synchronization with Forwarding! Bridges learn MC source addresses; during transients of the spanning tree, this should be disabled before a port is forwarding (root or designated) it goes through two intermediate states listening: wait for stabilization of ST (forwarding timer, 5 sec) learning: wait for addresses to be learnt (forwarding timer, 5 sec); all packets are forwarded by broadcast in this mode (no forwarding table is used) role/state event Port enabled Blocking by network management isabled Blocking Listening Learning Forwarding Port disabled isabled isabled isabled isabled (NM, failure) Port Selected Listening as Root or designated Port no longer Blocking Blocking Blocking root or designated Example. and root run STP procedure on new ports.. This triggers new BPUs sent to B and C. computes port p as new root. p at is set to listening state for 5 s. p at is set to learning state for 5 s $ topology change is fast (in this case), but forwarding not enabled immediately Forwarding Learning Forwarding timer expires 9 Source: CISCO RSTP White Paper 40

11 Host Cache Timers (f) Efficiency ()?! Timer used for validity of forwarding tables Two timer values are used long timer (5mn): normal case short timer = forwarding timer (5 sec): after spanning tree updates! Timer switching mechanism Bridge B detects change in ST -> maxlife := shorttimer how can bridges detect changes in ST"?! When one bridge senses a topology change topology change = one port changes out of or into blocking role bridge sends topology update BPU towards root (upstream bridges repeat BPU up to root) root forwards new config BPU with topology change flag set for a time duration = ForwardingTimer (5s) + Maxge timer (0s)S causes all bridges to use short timer value for caches until BPU from root received with topology change flag cleared i.e. broadcast from root! No loops, but paths may be not optimal ll frames go through the spanning tree 4 4 Efficiency ()?! elay to recover from a topology change is sec detect change : 0 to 0 s up to 0s if detected by absence of, or new, hello message, in average 0s immediate if detected by local physical link error message rebuild ST: s to a few seconds change port status to learning and forwarding: = * 5 s total: 0 s s = sec Q. If you want to improve the responsiveness of STP, which part would you improve in priority?. RSTP - Rapid ST protocol! Goal: fast reconfiguration Ethernet starts to be used as a public telecom service! Uses the same distributed Bellman-Ford algorithm changes are only to part (e) of STP: synchronization with packet forwarding Still a spanning tree parallel paths excluded!! Main improvements are fast transition to forwarding state with negotiation protocol instead of timeout fast flushing of forwarding tables after topology changes 4 44

12 Port Roles in RSTP port role is one of: root, designated, alternate, up root port = port towards root (same as STP) designated port = ports on which the bridge is designated for this LN (same as STP) Port that is not root nor designated is alternate, if the bridge is not designated for the LN connected to the port is up, if this is a second. parallel port to the same LN for which the bridge is already designated Other Port ttributes! port state discarding learning forwarding! edge / non edge! duplex (= point to point) / half duplex (= shared medium)! there are constraints on the combinations of role and state an alternate or up port is always in discarding state a root or designated can be any state alternate port = connects the bridge to root after topology update (alternative path to root) up port = connects LN to the spanning tree after topology update (alternative path to root for the LN) 45 4 Example Fast Transition to Forwarding L5 L9 R B B4 B R L4 R L L0 L8 U B L R L7 R L4 L L R - root ports - designated ports - alternative ports U - up ports! when a port is transitioned to designated or root it is put into the blocking state! a designated + blocking port starts the proposal agreement sequence in order to change its state to forwarding! RSTP produces the same ST as STP! L, L and are edge links, others not! L and L are half-duplex, others are full duplex new link opened ports blocking non edge downstream ports set to blocking upstream link set to forwarding 47 48

13 applies only to edge or full-duplex ports when a port is designated and blocking, it starts a proposal/agreement procedure by sending a proposal message when it receives an agreement in response, it is set to forwarding when a bridge receives a proposal message checks it is on root port starts a synchronization of all its ports, by blocking all designated, nonedge ports then it sets the root port to forwarding then it responds to the proposal message with an agreement Example of Fast Transition L5 L9 R B B4 B L4 R L4 R L L0 L8 U L B L R L7 R L! fails; the failure is detected immediately by physical layer alarms at B4 and B. What happens next? Solution Fast Flushing of Forwarding Tables Example! fter topology change (port becomes forwarding or blocking), STP puts cache timer to forwardingtimer (5s)! RSTP uses an active flushing all bridges send a BPU to neighbours every hellotimer ( = *s = 4s) ; L5 R L9 R B B4 B R L4 L L0 L4 L5 R L9 R B B4 B R L4 L L0 L4 when a non-edge port becomes forwarding (only; not when it becomes blocking) it sets the TC bit into the BPUs it will send for the next *hellotimer ( = *s = 4s) on all designated and root ports. This is done by setting the TC While timer on these ports. when a bridge receives a BPU with TC bit set, it flushes the forwarding tables on all designated and root ports except the one where received it sets the TC While timer on these ports causes all forwarding entries to be flushed in a few seconds flushing is thus piggyed into the ST re-computation BPUs Q: compare to STP 5 B L L8 R L7 R U L L L5 R L9 R B B4 B L4 R L4 L B L L8 R L7 R L0 U L L! fails; flushing propagates along spanning tree and lasts for a few seconds B L L8 R L7 R U L L L5 R L9 R B B4 B L4 R L4 L B L L8 R L7 R L0 U L L TC While time runs 5

14 Backup Ports RSTP - Rapid ST protocol L5 R L9 R B B4 B R L4 L L0 L8 B L R L7 R U L4 L L! Q: what are the main improvements over STP?! Q: what is not improved?! Q: o you know other technologies that recover from changes more quickly?! On shared medium LNs! failed port is immediately replaced by up port 5 54 Transparent Bridging (TB) nswers to questions! End systems ignore that there are transparent bridges bridge is transparent MC frames not changed by bridges frames not sent to bridge, but rather: bridge is promiscuous (listens to all frames)! Q. Is an Ethernet hub a bridge or a repeater? What s the difference?!. It can be either a bridge or a repeater. hub is a product name, not an architecture name. Modern hubs are bridges. Old ones are repeaters. The difference is: a repeater is a layer intermediate system (acts on bits) whereas a bridge is a layer intermediate system (acts on entire MC frames). lso: a bridge separates collision domains, a repeater does not hub hub hub 55 5

15 Can this method of learning addresses be extended to a network of bridges?! On this example, yes.! Q. How does B see the network?. B sees that,, B and Y are on port (B is transparent!) Its forwarding table is B C Y Z B port port B Y B C Z The method of learning does not work if there are loops in the topology! Q. What happens when send a frame to B? assume empty forwarding tables at the beginning. frame is sent by B to ports and. B learns that is on port. sends it to port. B sends it to ports and. B now learns that is on port. B sends frame to ports and etc the frame is multiplied a number of times. B receives several copies port port B port port port B port port B Y C Z Understanding the Specification cost = B8 B4 T Z B84 B90 B99 B9 Y cost = Examples of Path ttributes! Q. What are the attribute of the following paths? 0! 50! 40 [40, ] 0! 0! 0! 40 [0, 9] 0 0! 0! 0 [0, 7] 0! 50! 40! 0 [0, 4] The best path is the last root port blocking port designated port Forwarding Tables: B4 YZ T B84 YZT B9 ZT Y B8 YZT B90 ZT Y B99 ZT Y 59 0

16 Run the Centralized Bellman Ford lgorithm for Bridges on this Example p k (i): (format: (label, cost)) k \ i ,0 0,0 0,0 40,0 50,0 0,0 0, 0, 0, 0, 0,0 0, 0,7 0, 0, 0,0 0, 0,4 0, 0, i pred(i) Impact of Initial Conditions on the Bellman-Ford lgorithm for Bridges Q. write p k (i), pred(i) and draw the spanning tree, with initial conditions as shown. The dotted links do not exist in the current configuration. They existed before, and explain why nodes 0 and starts with these initial conditions.. no, it does not work. fter a few steps, all nodes believe the best label is 09, and start computing the best path towards 09. Then they start a count to infinity ( we are computing the usual distance to 09, which is infinite). The algorithm does not converge p k (i): (format: label, cost) k \ i ,0 0,0 09, 40,0 50,0 0,0 09,8 0, 0, 09, 09,9 09, 09,4 09,4 09, 09, 09,7 09, 09,9 09,0 4 09,8 09, 09, 09,0 09,0 5 09, 09,9 09, 09,0 09,0 09,0 09, 09, 09, 09, Example! Q. write p k (i), pred(i) and draw the spanning tree, with initial conditions as shown. The dotted link does not exist in the current configuration. It existed before, and explains why node 0 starts with these initial conditions.. The algorithm converges since the initial labels are not below the smallest one. 0 p k (i): (format: (label, cost)) 0 0 k \ i ,0 0,0 0, 40,0 50,0 0,0 0, 0, 0, 0, ,0 0, 0, 0, 0, 0,0 0, 0,4 0, 0, 4 0,0 0, 0,4 0, 0, The istributed Bellman-Ford lgorithm for Bridges may need to be reset! Like the centralized algorithm, the distributed algorithm is robust to changes in configuration as long as the node with the smallest label (called root bridge ) is still present and reachable from all bridges. If this is not true, the algorithm does not converge to a true value. It needs to be reset by some additional mechanism. Q. Compare to the classical distributed Bellman-Ford algorithm.. It does not need to be reset, since it always converges to the true value. 4

17 Topology Change That is Handled By Bellman-Ford Topology Change That is not Handled By Bellman- Ford! Q: If B4 dies, what happens?! : root monitoring at all bridges detect that B4 does not send a refresh message anymore T B84 B8 B4 B90 cost = Z all bridges start the STP procedure from fresh initial conditions and converge to a new spanning tree rooted at B8 Y B9 U B99 cost = 5 Efficiency ()?! elay to recover from a topology change is sec detect change : 0 to 0 s up to 0s if detected by absence of, or new, hello message, in average 0s immediate if detected by local physical link error message rebuild ST: s to a few seconds change port status to learning and forwarding: = * 5 s total: 0 s s = sec Q. If you want to improve the responsiveness of STP, which part would you improve in priority?. etection of changes and synchronization with forwarding B Example of Fast Transition L5 R L9 R B B4 B L4 R L4 L L0 L R L8 L7 R U L L B4 senses failure and clears state on port to ; sends new ST updates to neighbours; this converges to L5 being on the spanning tree 7 8

18 Example of Fast Transition Example of Fast Transition L5 R L9 R B B4 B R L4 L L0 L4 L5 R L9 R B B4 B R L4 L L0 L4 B L R L8 L7 R U L L B L R L8 L7 R U L L port on L5 at B starts proposal/agreement sequence; B4 blocks L9; L5 is opened port on L5 at B starts proposal/agreement sequence; B4 blocks L9; L5 is opened 9 70 Example of Fast Transition Fast Flushing of Forwarding Tables L5 R L9 R B B4 B R L4 L L0 L8 U B L R L7 R L4 L L! Q: STP propagates updates only from the root. Hello messages are generated by the root and relayed unchanged by the bridges. When a bridge receives an update, it simply sets the cache timer to a smaller value (5s) port on L9 at B4 starts proposal/agreement sequence; no port is blocked at B; L9 is opened. 7 7

19 RSTP - Rapid ST protocol! Q: fast move to forwarding state after topology changes! Q: what is not improved? flushing stale forwarding entries takes several seconds, not a large improvement over STP parallel paths are not used, except as up! Q: self healing double rings 7