Classifications

• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L45/00—Routing or path finding of packets in data switching networks
  • H04L45/28—Routing or path finding of packets in data switching networks using route fault recovery
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L12/00—Data switching networks
  • H04L12/02—Details
  • H04L12/16—Arrangements for providing special services to substations
  • H04L12/18—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast
  • H04L12/1854—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast with non-centralised forwarding system, e.g. chaincast
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L12/00—Data switching networks
  • H04L12/02—Details
  • H04L12/16—Arrangements for providing special services to substations
  • H04L12/18—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast
  • H04L12/1863—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast comprising mechanisms for improved reliability, e.g. status reports
  • H04L12/1868—Measures taken after transmission, e.g. acknowledgments
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
  • H04L41/06—Management of faults, events, alarms or notifications
  • H04L41/0654—Management of faults, events, alarms or notifications using network fault recovery
  • H04L41/0668—Management of faults, events, alarms or notifications using network fault recovery by dynamic selection of recovery network elements, e.g. replacement by the most appropriate element after failure
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L43/00—Arrangements for monitoring or testing data switching networks
  • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
  • H04L43/0805—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters by checking availability
  • H04L43/0811—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters by checking availability by checking connectivity
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L45/00—Routing or path finding of packets in data switching networks
  • H04L45/22—Alternate routing

Definitions

• the present invention generally relates to media streaming, and especially to unicast or multicast streaming and further especially to handling reconnections in a transmission tree.
• Data streaming e.g. unicast or multicast streaming might be performed from any source to a plurality of communication nodes that are interconnected tree-like so that every node is coupled, either via a direct interconnection or via interconnection involving one or a plurality of intermediate nodes, to the source.
• One problem is to handle faults within the tree; e.g. a link failure or a node failure.
• Protocol Independent Multicast Sparse Mode is a well known and commonly applied protocol for building up and maintaining multicast trees in IP networks. This solution uses a single tree for forwarding packets to routers with hosts (destinations in the sequel) wanting to receive the content. PIM-SM is called “protocol independent” because it can use route information that any routing protocol enters into the multicast Routing Information Base.
• a router When a router wants to join or leave a multicast group, it can do it using PIM-SM using simple unicast forwarding.
• a node When a node wants to join to a multicast tree using PIM- SM, it sends a JOIN message back towards the source (or towards the rendezvous point for shared tree; in the sequel we will not distinguish between these two anymore) More precisely, the last hop router of the destination may send some JOIN messages to the source (source-based tree) or to the rendezvous point (shared tree).
• the JOIN packet is routed along a path determined by Multicast RIB (MRIB).
• MRIB Multicast RIB
• the MRIB is used to determine the next-hop neighbor to which any PIM Join/Prune message is sent.
• JOIN is routed and processed hop-by-hop until a node already receiving the traffic is reached. All routers along this path process the JOIN message and install/update multicast routing state (e.g. adding the incoming interface to the outgoing interface list). Data flows along the reverse path of the JOIN messages.
• PIM JOIN packets are forwarded along the shortest path to the rendezvous point or to the source, which may differ from the shortest downstream path in the case of asymmetric link costs.
• multicast streams established with PIM potentially use suboptimal paths downstream (e.g. reverse shortest paths). Later, multicast packets will be forwarded along this path. Similarly, a destination wanting to leave the group sends a PRUNE packet up the tree. More detailed information about PIM SM can be drawn e.g. from the IETF document RFC4601 1 .
• MLDP Multicast Label Distribution Protocol
• MLDP Label Map message sent from the egress points of the tree towards the root of the tree (the root in the MPLS network).
• the effect of the MLDP Label Map message is similar to the PIM Join message as discussed above.
• the MLDP Label Map message also goes upstream and immediately installs the MPLS labels to be used downstream.
• PIM-SM depends on unicast routing such that if the routing fails, it must wait for the unicast routing to recover, thus making the convergence relatively slow. Since PIM-SM is commonly for building up paths for real-time traffic (e.g. for IPTV), this slow convergence can be a serious drawback. The same is true for MLDP.
• a plurality of nodes (in the following also being referred to as communication or network nodes) of a communication network are connected to form a distribution tree such that data traffic is forwarded from a first node (e.g. a source node) to each of the plurality of further nodes.
• a first node e.g. a source node
• a first node detects a data traffic interruption with respect to second node e.g. by receiving a failure notification from the second node or by detection a connection loss to that node, the node determines if the data traffic is still received at this node e.g. if it is not an ancestor node with respect to the traffic flow). If yes, the first node forwards the data traffic to the second node; otherwise it notifies a thirds node about the connection failure.
• the third node If the third node is receiving the data traffic (over another tree branch), it forwards the data traffic to the first node e.g. by unblocking or activating an interconnection normally blocked or inactivated. If (e.g. as a consequence of the failure) the third node also does not receive the data traffic (e.g. if the third node is situated on the same tree branch), the third node notifies a fourth node, e.g. by forwarding the received message to the fourth node. The above procedure might be repeated in the fourth node and further nodes, until a last notified node is a node still receiving the data traffic. Data traffic might be forwarded backwards to every node in the chain of notifying nodes.
• the nodes of a distribution tree keeps stored each an alternative node to which it can connect for receiving the traffic in the case of a failure with effect to the main connection.
• the recipient (the alternate node) of an failure notification (e.g. an activation packet) is actually receiving the stream from a node that sent the notification; the recipient forwards the activation request further to its own alternate. Otherwise the node starts forwarding the data packets towards the requesting node. It might be noted that one or a plurality of further nodes might be outside the distribution tree adapted to forward notification packets.
• an failure notification e.g. an activation packet
• the present invention also concerns computer programs comprising portions of software codes in order to implement the method as described above when operated by a respective processing unit of a user device and a recipient device.
• the computer program can be stored on a computer readable medium.
• the computer-readable medium can be a permanent or rewritable memory within one of the nodes or located externally.
• the respective computer program can be also transferred to the nodes for example via a cable or a wireless link as a sequence of signals.
• Fig. 1 shows an illustrative example for a multicast data distribution tree
• Fig. 2 shows an example for a node failure problem
• Fig. 3 shows an illustrative sequence of steps being performed in a network according to Fig. 1 or Fig. 2 after an occurrence of a transmission failure
• Fig. 4a, b, c each show an exemplary section out of the network of Fig. 1 , and are illustrative of different stages for reconnection over a alternate link at a failure of a main link.
• a source node 1 1 and a plurality of (communication) nodes 1 1- 19 are interconnected to form a data distribution tree such that each of the plurality of nodes 1 1- 19 receives data either directly from the source node 10 or indirectly over one or a plurality of interconnected nodes.
• each of the recipient nodes 1 1 -19 has one main (or initially activated) link or connection (drafted as solid arcs) c01 , c09, c12, c13, c23, c65, c54, c97, and c98 to an ancestor node in order to receive the data.
• a plurality of alternate (or initially non-activated) links or connections c32, c34, c43, c56, c61 , c57, c75, c79, c97, and c78 (drafted as dotted arcs) between each a pair of nodes are provided.
• the main connections are forming a distribution tree that might be regarded as primary tree, and the alternate connections might be regarded to forming a protection forest.
• the alternate connections might be selected in such a way, that when a failure occurs at any point of the network, it is always possible to patch the original tree by using some of the alternate connections. If it is e.g. supposed that a certain connection (e.g. first connection c01 ) between first (recipient) node 1 1 and the source node 10 goes down, first node 1 1 , after loosing its connection, tries to rejoin to its alternate, i.e. to sixth node 16 in this example. However, as this node received the traffic from the first node 1 1 , it needs to use its alternate as well, so it reconnects to fifth node 15.
• a certain connection e.g. first connection c01
• this node reconnects to seventh node 17, which is the first node in the chain that is not an ancestor of the fifth node 15 or of the first node 1 1 , so it is able to inject the traffic back to the lost component of the tree comprising the nodes 1 1 , 12, 13, 14, 15 and 16.
• Some applications may require small fail-over times. In these cases, reconnecting to the remaining part of the tree should be as fast as possible. Therefore, when a node detects the loss of connection, it sends out an activation packet immediately to its alternate.
• Alternate connections can be thought of as inactive forwarding state: e.g. they are installed in the multicast FIB entries but are marked as blocked until an activation packet unblocks or activates them.
• the inactive forwarding states might have been, in this case, installed by prior tree-building mechanism, with an extension that these are now marked as inactive (blocked forwarding state).
• the recipient of an activation packet was actually receiving the multicast stream from node that sent the activation, the recipient is suspected to immediately forward the activation packet further to its own alternate (e.g. sixth node 16 receiving the indication must forward it to fifth node 15). Otherwise the node just needs to start forwarding multicast data packets towards the requesting node by unblocking the forwarding entry (e.g. after receiving an activation message, seventh node 17 starts sending data packets to fifth node 15, which in turn sends these packets also to sixth node 16 that forwards these packets to first node 1 1 .
• the approach described above may also be able to mitigate node failures. E.g. in a case that not a link, but an associated node fails (e.g.
• one or a plurality of immediate affected nodes down in the tree e.g. node 12 and node 16
• these nodes will send out a failure notification each to their alternate node.
• Sixth node 16 will reconnect as previously. Additionally, now, second node 12 will reconnect as well to third node 13, so two trees are needed to patch this failure.
• Fig. 4a - 4c exemplarily illustrate a summary of the above-said.
• a part of the distribution tree of Fig. 1 comprising nodes 1 1 , 16, 14, H and 17 is shown.
• Node 1 1 is connected to node 16 over connection c16, node 16 is connected to node
• Fig. 4a shows a first state with activated connections c16, c65, c54, and non-activated connection c75 such that the data is distributed from node 1 1 to node 16, from this node to node 15, and from this node to node 14. Further the same data is sent to node 17 over a different path.
• Fig. 4b illustrates an exemplary failure at the link c16 between node 1 1 and node 16 resulting in that nodes 16, 15 and 14 will not receive the data anymore.
• Fig. 3 illustrates a basic sequence of steps S1 , S2, S31 , and S32:
• a first node e.g. node 16, 15 or 17, detects a data traffic interruption with respect to second node e.g. nodes 1 1 , 16 or 15.
• this node detects if the data traffic is still received at the first node (i.e. if it is not an ancestor node with respect to the traffic flow). If yes, in first alternative step S31 the first node forwards the data traffic to the second node; otherwise in a second alternative step S32, it notifies a thirds node about the connection failure, wherein the thirds node might repeat the previous steps.
• the network is a packet switched network, where multicast is realized by virtual circuit switched paths.
• each node might know the complete topology (e.g. application of a link state routing protocol in the network), and the multicast trees are built up by reverse shortest paths.
• the nodes do not necessarily know the exact multicast tree, i.e. they do not necessarily know which nodes are in the group (if there are multiple shortest paths, it might be impossible to predict, which one will be selected).
• packets are forwarded from the source to the destination, which is a correct assumption for the source based tree mode of PIM-SM (for the shared tree of PIM-SM, the term "source" is the rendezvous point in the sequel).
• a work tree is built up that is the "primary" tree. Further, each node computes an alternate parent, which is either one of its children in the multicast tree or one of its neighbors, which is neither a successor nor an ancestor (here an ancestor is a node closer to the root along the multicast tree). Alternates might be pre-computed, so when a failure occurs, the detecting node immediately sends out an activation packet, which enables its alternate to send the multicast traffic to that node.
• a multicast tree to forward packets is built by PIM (by way of example, the direction from the root to the destinations along the tree is referred to as down direction, and the opposite direction is referred to as up direction). If there is a failure splitting the forwarding tree into multiple components, tree parts need to be patched somehow and the lost component(s) are to be reconnected back to the remaining tree. In exemplary embodiments of the invention a mechanism is described, which can realize this reconnecting capability rapidly after a failure.
• Reconnecting to the remaining part of the tree may happen with regular tree building mechanisms, like PIM Join or its equivalent in mLDP (Label Map message).
• the fail- over time is limited by the performance how nodes process the regular tree-building packets, which is typically done in the control plane.
• the above-described technique is applicable to a broad range of telecommunication networks, e.g. where (virtual) circuit switched multicast paths are applied, such as in IPv4 and v6 (e.g. using PIM) and MPLS networks (e.g. using MLDP or RSVP-TE).
• IPv4 and v6 e.g. using PIM
• MPLS networks e.g. using MLDP or RSVP-TE.
• Fig.2 describes an example, wherein as a result of a single failure, the network is split into not only two, but more components.
• Fig.3 exemplarily shows source node 10 and a plurality of receiving codes 21 , 22, 23, 24, 25, 26 and 27.
• a failure of node 21 splits the network into three components T1 by way of example comprising two nodes 23 and 24, T2 by way of example comprising three nodes 25, 26, and 27 and T3 by way of example comprising node 22.
• the first component T1 rooted at node 23 needs to get to the one rooted at node 25 before getting out from the failed area rooted at node 21 .
• node 24 of the first component T1 has a non-activated connection to node 26 of the second component T2. Since the parent of node 24 cannot be node 26 at this point, so node 26 will not recognize the loss of connection when node 24 activates it as an alternate. However, this is a similar situation of the case as discussed under Fig.1 , when node 1 1 was failing. Since the second component T2 is rooted at node 25, this note after detecting the failure will restore the connection for node 26.
• all the nodes not in the multicast tree but selected as an alternate should select two alternates as well.
• the primary among these alternates is the neighbour, which would be parent, if the node was in the multicast tree.
• the secondary alternate is the neighbour, which would be the alternate, if the node was in the multicast tree. All the nodes need to join to both alternates as described previously, which means that both of these neighbours will do two alternate joins as well, if they are not in the multicast tree.
• a node not being in the multicast tree receives an activation packet, it needs to select one of its alternates, and forward the activation packet to that neighbour.
• this neighbor is always the primary alternate, except it is not available (the node needs to detect the failure of its primary alternate, (e.g. by means of bidirectional forwarding protection, as e.g. described in a document of D. Katz, and D. Ward, titled “Bidirectional forwarding detection", IETF RFC5880, June 2010) or if the sender of the activation packet is the primary alternate itself. In those two cases activation packet must be forwarded to the secondary alternate.
• rejoining is exemplarily done in the same way as in the case when all the nodes were inside the multicast tree. If it is possible, nodes outside the tree will build up the branch of the tree containing them. Otherwise, they build up their alternate path leading out from their failing component.
• a node can be in multiple multicast groups simultaneously, which means that for this case, keeping up one (or two if the node is not in the multicast tree) alternate per protected multicast groups is needed.
• This activation message is preferably a simple packet, which can be processed at the data plane of the router (in order to reduce reaction time).
• the packet must describe which multicast tree went down either by using some special destination address, or storing this information in the packet. Moreover, it is possible that the same alternate is used for more than one multicast group. In that case the activation packet may contain all of these groups.
• a node receiving the activation packet should immediately start forwarding multicast traffic to the sender of the packet.
• multicast with some blocking possibility; when it is needed, we the forwarding plane can simply remove blocking.
• the alternate uses its own alternate in order to restore traffic flow, either when the sender of the activation packet is the parent of the receiver or when the receiver is not in the multicast tree. In this case the activation packet must be forwarded towards the alternate of the receiver.
• SRLG Shared Risk Link Group
• SRLGs resulted by some LANs in the network can be protected in a similar way; such LANs should be presented with some virtual nodes in the graph of the network, and we need to avoid them, as they were ordinary routers.
• a node wanting to join, leave or upkeep a multicast group supported by protection forest needs to do exactly the same tasks as currently (e.g. needs to periodically send out some JOIN packets for PIM). Moreover, now each node has some alternate for each of the multicast groups to which it has joined. Since alternates need to prepare to send out the traffic to some of their interfaces immediately, when some activation packet is received, an alternate needs to know that it has been selected. Therefore alternate selection can be realized with some ALTERNATE JOIN messages.
• ALTERNATE JOIN messages Being an alternate might be a soft state (since no communication network is completely reliable), therefore such ALTERNATE JOIN messages is preferably sent out periodically in order to keep up this state. If such packet is not received for a certain period of time, this state is removed.
• a mechanism might be needed to notify neighbors, when a node wants to leave a multicast group. In that case, the node can either simply stop sending ALTERNATE JOIN messages, or send out some special ALTERNATE PRUNE message to its previous alternate.
• mLDP can be used, which can be regarded as conceptually similar to PIM in IP.
• the mLDP Label Map message is similar to PIM Join message, i.e. goes upstream and installs the labels to be used by the data traffic downstream. So, for mLDP an "Alternate Label Map" message will be required, which can be a Label Map message with a flag set.
• a failure detection mechanism between each nodes sending multicast traffic to each other can be realized e.g. by some hardware element (e.g. the loss of voltage can be detected) or by bidirectional forwarding detection (BFD) in the above-cited document IETF RFC5880.
• some hardware element e.g. the loss of voltage can be detected
• BFD bidirectional forwarding detection
• the protection forest When the failure has proven itself to be permanent, the protection forest should be reconfigured in order to prepare for another failure. In that case, nodes connected to their alternates should finish sending ALTERNATE JOIN messages and should join to their alternate as a parent. In this way the patched tree can be fixed. Later this tree can be optimized (this can be regarded as a responsibility of the protected multicast algorithm; e.g. PIM can rejoin to the best next hops using some JOIN packets).
• reconfiguration with protection forest is much faster in the case of any single failure than in the case of normal PIM- SM or mLDP (an activation of the backup path might fall far below the 50ms convergence limit of fast reroute).
• an implementation of the proposed mechanism in the data plane might be rather simple by just removing some blocking, if a special packet was received.

Landscapes

• Engineering & Computer Science (AREA)
• Computer Networks & Wireless Communication (AREA)
• Signal Processing (AREA)
• Environmental & Geological Engineering (AREA)
• Data Exchanges In Wide-Area Networks (AREA)

Priority Applications (3)

Applications Claiming Priority (2)

Publications (1)

Family

ID=47146352

Family Applications (1)

Country Status (4)

Cited By (1)

Families Citing this family (3)

Family Cites Families (5)

• 2012
  • 2012-10-17 IN IN845KON2014 patent/IN2014KN00845A/en unknown
  • 2012-10-17 EP EP12783548.6A patent/EP2769504B1/en active Active
  • 2012-10-17 US US14/352,341 patent/US9509557B2/en not_active Expired - Fee Related
  • 2012-10-17 WO PCT/EP2012/070600 patent/WO2013057158A1/en not_active Ceased

Non-Patent Citations (4)

Also Published As

Similar Documents

Legal Events