CA2162491C - Method for efficient aggregation of link metrics - Google Patents
Method for efficient aggregation of link metrics Download PDFInfo
- Publication number
- CA2162491C CA2162491C CA002162491A CA2162491A CA2162491C CA 2162491 C CA2162491 C CA 2162491C CA 002162491 A CA002162491 A CA 002162491A CA 2162491 A CA2162491 A CA 2162491A CA 2162491 C CA2162491 C CA 2162491C
- Authority
- CA
- Canada
- Prior art keywords
- link
- virtual
- spanning tree
- delay
- bandwidth
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Lifetime
Links
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
- H04L45/02—Topology update or discovery
-
- 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
- H04L45/02—Topology update or discovery
- H04L45/03—Topology update or discovery by updating link state protocols
-
- 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
- H04L45/44—Distributed routing
-
- 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
- H04L45/48—Routing tree calculation
Abstract
The unique method of the present invention provides efficient aggregation of link metrics for a sub network in a communication network having a plurality of interconnected subnetworks that consist of nodes and links. Thus, link metric values are summarized on a full-mesh topology and encoded by a spanning tree topology.
The method encodes the link metric information contained in full-mesh topology with a minimum compromise in the capability to recover it.
The method encodes the link metric information contained in full-mesh topology with a minimum compromise in the capability to recover it.
Description
2~62~91 METHOD FOR EFFICIENT AGGREGATION OF LINK
METRICS
s Field of the Invention This invention relates generally to link state parameters in a communication network, and more ~o particularly to efficient aggregation of link states for a subnetwork in a communication network.
Background is In a communication network consisting of nodes representing switching systems. and links representing transmission facilities between pairs of nodes, the network topology, i.e., the collection of all link state information, is maintained by each node that is responsible ao for path computation for establishing communication between nodes. Routing protocols, which are used to select paths, may be broadly divided into two classes: link state protocols and distance vector protocols. Link state protocols require each node to maintain the states of all Zs relevant links. Distance vector protocols require each node to maintain the state of the shortest path from itself to each relevant destination.
Link metrics are quantitative link state parameters so that are associated with the quality of information transfer and are used in path computation. There are two kinds of link metrics: non-additive and additive link metrics. A non-additive link mEaric generically takes the form of "bandwidth". An additive link metric generically ss takes the form of "delay". The generic path selection 2 21~24~1 problem is to determine a path according to a given routing objective, such that the bandwidth of each link on the selected path is greater than a given bandwidth threshold, and that the delay along the selected path is smaller than a s given delay threshold. There are two common routing objectives in the prior art: maximum bandwidth and minimum delay.
Path metrics are derived from link metrics. For a io non-additive link metric such <~s bandwidth, the metric value associated with a path is the minimum metric value among all the links along the path. For an additive link metric such as delay, the metric value associated with a path is the sum of the metric values of the links along the is path.
The topology of the network is kept in a topology database attached to each nodle that is responsible for maintaining and using such information. Due to changes in 2o the network, the topology must be updated from time to time via a topology broadcast mechanism. A topology broadcast is an event executed' by a node such that a message containing link state information is advertised or distributed to all other nodes in the network.
There are two general approaches in the prior art to address topology update complexity. In one approach, the prior art reduces complexity by limiting the handling of the topology to a subset of the nodes in the network. In so another approach, the prior art aggregates the link states in each subnetwork so that nodes outside the subnetwork need only maintain partial information on its topology.
Typically, a representative node is elected from the nodes in the subnetwork to be responsible for aggregating and
METRICS
s Field of the Invention This invention relates generally to link state parameters in a communication network, and more ~o particularly to efficient aggregation of link states for a subnetwork in a communication network.
Background is In a communication network consisting of nodes representing switching systems. and links representing transmission facilities between pairs of nodes, the network topology, i.e., the collection of all link state information, is maintained by each node that is responsible ao for path computation for establishing communication between nodes. Routing protocols, which are used to select paths, may be broadly divided into two classes: link state protocols and distance vector protocols. Link state protocols require each node to maintain the states of all Zs relevant links. Distance vector protocols require each node to maintain the state of the shortest path from itself to each relevant destination.
Link metrics are quantitative link state parameters so that are associated with the quality of information transfer and are used in path computation. There are two kinds of link metrics: non-additive and additive link metrics. A non-additive link mEaric generically takes the form of "bandwidth". An additive link metric generically ss takes the form of "delay". The generic path selection 2 21~24~1 problem is to determine a path according to a given routing objective, such that the bandwidth of each link on the selected path is greater than a given bandwidth threshold, and that the delay along the selected path is smaller than a s given delay threshold. There are two common routing objectives in the prior art: maximum bandwidth and minimum delay.
Path metrics are derived from link metrics. For a io non-additive link metric such <~s bandwidth, the metric value associated with a path is the minimum metric value among all the links along the path. For an additive link metric such as delay, the metric value associated with a path is the sum of the metric values of the links along the is path.
The topology of the network is kept in a topology database attached to each nodle that is responsible for maintaining and using such information. Due to changes in 2o the network, the topology must be updated from time to time via a topology broadcast mechanism. A topology broadcast is an event executed' by a node such that a message containing link state information is advertised or distributed to all other nodes in the network.
There are two general approaches in the prior art to address topology update complexity. In one approach, the prior art reduces complexity by limiting the handling of the topology to a subset of the nodes in the network. In so another approach, the prior art aggregates the link states in each subnetwork so that nodes outside the subnetwork need only maintain partial information on its topology.
Typically, a representative node is elected from the nodes in the subnetwork to be responsible for aggregating and
3 216~4~1 advertising the subnetwork link metrics to nodes outside the subnetwork.
The prior art that makes no compromise in the s amount of link state information divides all nodes into two types: (i) master nodes with extended memory and computing capabilities, and (ii) subordinate nodes with limited memory and computing capabilities. A topology database is attached to each master node and is used by ~o the node for path computation. 'the subordinate nodes rely on the master nodes for topology maintenance and path computation.
In an extreme approach for link metric aggregation, ~s the entire subnetwork is repre:>ented by a pseudonode.
Such an approach can be traced back to the IS-IS, i.e., Intermediate System-to-Intermediate System, routing protocol. In one version of this approach, all internal link metrics of the subnetwork are reduced to a node metric for Zo the pseudonode by taking worst cases. For an additive link metric, a popular assignment of the corresponding node metric is the "diameter" or the length of the longest shortest path between any pair of nodes in the subnetwork.
For a non-additive link metric, a popular assignment of the Zs corresponding node metric is the bandwidth of the smallest maximum-bandwidth path between any pair of nodes in the subnetwork. Another version of the approach is to let the pseudonode to represent the "middle" of the subnetwork, and assign an appropriate metric, by taking 3o averages, to each of the links connecting the psuedonode and the exposed nodes. A variation of the above versions makes use of a diameter variance metric in addition to the diameter metric to reflect deviation from a symmetric topology.
216~~91 In the OSPF, i.e., the Open Shortest Path First, routing protocol, summary link states are used for inter-subnetwork routing. The summary links are essentially s virtual links to external nodes. From the point of view of a subnetwork, there are internal links that connect the internal nodes together, and there are summary links that emanate from the subnetwork to external nodes. OSPF is designed to be a link state routing protocol.
io In the prior art, a network is clustered into a number of levels, and each node has knowledge of only the shortest paths within its own lowest level cluster and a single shortest path to each supercluster. Path selection is is based on a distance-vector method, where a routing table indicating the best next hop for each intended destination is maintained by each node.
Also in the prior art that aggregates link states, the 20 original topology of the subnetwork is reduced to a smaller topology consisting of a subset of the nodes, namely exposed nodes, in the subnetwork. This method advertises a full-mesh virtual topology whose nodes are the exposed nodes and whose links represent paths in the original Zs topology. The paths are determined with respect to a given routing objective. The links in the virtual topology are known as virtual links. This approach is well suited to a hierarchical routing architecture.
3o In other prior art, a different problem is solved. A
heuristic that starts by computing a full-mesh shortest-path network consisting of a subset of the nodes in a given original network and then reducing the full-mesh network to a minimum spanning tree is used for routing to multiple destinations in a computer network, where the object is to minimize the total link weight on the multicast tree that connects the destination nodes together. There are two common spanning tree algorithms known in the prior art:
s maximum spanning tree algorithm and minimum spanning tree algorithm.
Although the approaches tlhat advertise the topology in terms of a pseudonode, i.e., variationsthe approach of in io IS-IS, offer the greatest reduction of advertised information, these approaches typically do not supply enough information for efficient routing.Moreover, the version that handles worst casE~s cannot reflect the link state information associated with an asymmetric is subnetwork topology. The version that takes averages cannot offer any performance guarantee.
The version that uses a diameter variance metric i s still under investigation. When a subset of the nodes in the subnetwork are exposed, all versions do apply.
not The approach that advertises the full-mesh representation of the subnetwork requires advertisement of order M2 pieces of link state information when the number of exposed nodes is M. For a subnetwork whose 2s nodes are not very densely connected, the amount of link state information to be advertised may well exceed that contained in the original topology of the subnetwork when M is sufficiently large. Such explosion of link state information contradicts the purpose of link metric so aggregation, i.e., efficiency in routing.
The summary link approach in OSPF does not capture enough link state information for efficient routing that is subject to quality of service constraints. Specifically, it 6 ~~s~~~1 does not offer the flexibility to pick and choose link resources that best meet the quality of service requirements for a connection.
s Efficient aggregation of link metrics should provide more efficient routing through the communication network.
Hence, there is a need for a method for efficient aggregation of link metrics such that the number of link metrics to be advertised is minimized without appreciably io compromising the information contained in the link metrics.
Brief Descriptions of the Drawings is FIG. 1 is a representation of a topology of a subnetwork for utilizing the method of the present invention.
ao FIG. 2 is a representation of a full-mesh topology of a subnetwork for utilizing the method of the present invention.
FIG. 3 is a representation o~f a spanning tree topology Zs of a subnetwork for utilizing the method of the present invention.
FIG. 4 is a representation of exemplary bandwidth and delay metrics in a subnetwork for utilizing the method 30 of the present invention.
FIG. 5 is a representation of exemplary bandwidth and delay metrics for a full-mesh topology derived from a... ~ ~~6~~91 maximum bandwidth paths utilizing the method of the present invention.
FIG. 6 is a representation of exemplary metrics for a s maximum bandwidth spanning tree in accordance with the method of the present invention.
FIG. 7 is a representation of exemplary metrics for a maximum delay spanning tree in accordance with the to method of the present invention.
FIG. 8 is a representation of exemplary metrics advertised for bandwidth and delay in accordance with the method of the present invention.
FIG. 9 is a representation ~of exemplary estimates for a full-mesh topology in accordance with the method of the present invention.
ao FIG. 10 is a flow chart of one embodiment of steps in accordance with the method of the present invention, which provides efficient aggregation of link metrics for a subnetwork in a communication network having a plurality of interconnected subnetworks that consist of nodes and 2s links.
FIG. 11 is a flow chart of one embodiment of steps for aggregation of link metrics fior bandwidth maximization utilizing the Maximum Bandwidth Link Metric so Aggregation Heuristic in accordance with the method of the present invention.
FIG. 12 is a flow chart of one embodiment of steps for aggregation of link metrics for delay minimization 2~.6~491 g utilizing the Minimum Delay Link Metric Aggregation Heuristic in accordance with the method of the present invention.
s FIG. 13 is a flow chart of a,n embodiment of steps in accordance with the method of the present invention.
FIG. 14 is a flow chart of one embodiment of, in the step of assigning link metric valn,~es to the virtual links in io accordance with the present invention, selecting a virtual link, on the full-mesh virtual topology, whose metric value is to be estimated from the corrE~sponding spanning tree.
~s Detailed Description of a Preferred Embodiment In a large internetworking environment consisting of many subnetworks, aggregation of link metrics would be advantageous, in particular for two reasons. First, the Zo amount of link state information must be compressed to reduce the topology update complexity which is proportional to the number of links. Second, the actual topology of a subnetwork may have to be hidden for security reasons. The method of the present invention is as applicable to link state routing protocols, and provides an efficient solution to the problem of aggregating link states, specifically link metrics, within a subnetwork where a predetermined subset of the nodes are exposed, i.e., to be retained in the process of aggregation. This 3o method requires significantlly less communication bandwidth for topology updates than the conventional method where every node in tlhe network is involved in maintaining and updating the entire topology of the network. The method of the present invention may also be applied recursively to a network with hierarchical structu res.
The purpose of link metric aggregation is to s represent the original topology in the subnetwork by a virtual topology that consists of only the exposed nodes and the smallest number of virtual links that can adequately capture the internal link metrics of the subnetwork for the purpose of path selection based on a io given routing objective. There may be one such virtual topology for each link metric. The virtual topologies do not have to be the same since only the link metric information derived from them is. of importance.
is For example, consider aggregating link metrics for a subnetwork with N connected nodes, M of which are exposed, or retained in the aggrecaation, where M < N, where M and N are positive integers. F=IG. 1, numeral 100, shows an exemplary subnetwork (102) whose exposed nodes (A, B, ao C, D) are shaded. For M exposed nodes, the largest number of virtual links that connect them together, since no parallel links are allowed, is M(M-1 )/2, i.e., order M2, and such a fully connected topology, FIG. 2, numeral 200, is known as a full-mesh topology (fully interconnected A-B-as C-D; 202). The smallest number of virtual links that connect all the exposed nodes i:ogether is (M-1 ), and any connected virtual topology with (M-1 ) virtual links belongs to the family of topologies known as "trees". Specifically, since all M exposed nodes are o~n the tree in this case, the so tree, shown in FIG. 3, numeral 300, is known as a "spanning tree". A tree topology consists of no loops. Also, there is a unique path connecting each pair of nodes on the spanning tree.
Where topology update complexity is not an issue, one would clearly want to advertise all the link metric information contained in the full-mesh virtual topology, thereby obtaining the best, i.e.,, most complete, network s topology without permitting parallel links. The representation of the original subnetwork topology by a full-mesh topology connecting a predetermined set of exposed nodes is referred to as a complete summarization of the subnetwork topology. Where selected, the io information contained in the complete summarization is compressed by means of encoding. The method of the present invention encodes they link metric information contained in the full-mesh topology, together with minimum compromise in the capability to recover it. The is order M2 pieces of link metric information are compressed to order M pieces. Given this encoded link metric information, one can obtain (M-1 ) pieces of the link metric information with complete accuracy and derive reasonable estimates of the remaining pieces in the form of lower ao bounds for bandwidth metrics and upper bounds for delay metrics. In some cases, the Estimates are so accurate that they are equal to the exact metric values. In this way the number of link metrics to be advertised is minimized without appreciably compromising the information as contained in the link metrics.
The aggregation and distrik~ution of link metrics for a subnetwork are typically executed by a representative node elected from the nodes in the subnetwork. This node 3o is referred to as the subnetwork leader. The link metric advertisement is to be receivE;d by nodes outside the subnetwork. These nodes are referred to as external nodes.
Upon receiving the advertisement, each external node is w :' 2~ g249't ~~
responsible for decoding or estimating the values of the aggregated link metrics of the subnetwork.
FIG. 10, numeral 1000, is a flow chart of one s embodiment of steps in accordance with the method of the present invention, which provides efficient aggregation of link metrics for a subnetwork in a communication network, comprising the steps of: A) selE:cting, by a predetermined subnetwork leader, a subset of the nodes in the subnetwork io to be exposed nodes (1002); B) for each link metric, generating, by the subnetwork leader, a spanning tree virtual topology connecting exposed nodes (1004); and C) assigning link metric values to~ the virtual links on the spanning tree and advertising the metric values in a is topology broadcast by the subnetwork leader (1006).
Where selected, one may add, after step (A) a step of reducing, by the subnetwork leader, an original subnetwork topology, stored in a topology database of the subnetwork leader, to a full-mesh virtual topology wherein each Zo virtual link connecting a pair of .exposed nodes on the full-mesh virtual topology is derived utilizing a path selection optimization (1008). The advertised link metric values in step (D) typically include a number of predetermined specific link metric values for replacing selected Zs estimates derived from each spanning tree.
Thus, the original topology of the communication subnetwork is reduced by the subnetwork leader to a full-mesh virtual topology. Connecting a set of M predetermined 3o exposed nodes, M a positive integer, such that the metric values on each virtual link on the full-mesh virtual topology are derived from a path selection optimization, e.g., a maximum bandwidth path or a minimum delay path.
Then, for each link metric, the subnetwork leader ~' ~2 :2162491 generates a spanning tree viri:ual topology from the full-mesh virtual topology for encoding the metric values on the full-mesh topology. ThE~ subnetwork leader then advertises the encoded link metric values, i.e., those s assigned to the virtual links on the spanning trees, to the external nodes via a topology broadcast. For each link metric, a full-mesh topology connecting the M exposed nodes is derived, by each extE~rnal node, by means of a decoding technique, set forth below, from the spanning io tree topology such that the link metric value associated with each virtual link on the derived full-mesh topology is an estimate of the link metric value associated with the corresponding virtual link in the original full-mesh representation of the subnetwork topology.
Although generally the link metric values associated with a link are substantially symmetrical in both directions of a link, the method of the present invention is also applicable where link metric values are not ao symmetrical. In the case of the generic link metrics bandwidth and delay, the link bandwidth may be taken to be the minimum bandwidth between the two directions, and the delay to be the maximum delay between the two directions. In the present invention, an estimate for Zs bandwidth is a lower bound, and' where it is a lower bound on the minimum bandwidth between the two directions of a link, the bandwidth estimation is the lower bound on bandwidth for both directions. I~n the present invention, an estimate for delay is an upper bound, and where it is an 3o upper bound on the maximum delay between the two directions, the estimate for delay is the upper bound on delay for both directions.
The present invention provides for selectively 3s utilizing, for example, one of two types of path selection ~3 zlsz4~~
optimization: bandwidth maximization and delay minimization. A path for the bandwidth maximization is determined utilizing a maximum bandwidth path algorithm.
A path for the delay minimization is determined utilizing a s minimum delay path algorithm. The maximum bandwidth path algorithm is a straightforward variation of the shortest path algorithm. The present invention also utilizes a graph algorithm , the Maximum Spanning Tree algorithm, and its equivalent version Minimum Spanning io Tree. The Maximum/Minimum Spanning Tree algorithm is used to determine a spanning tree with the maximum/minimum sum of link weights.
The present invention combines the selected path is optimizationand spanning algorii:hms in a unique manner to provide a new, efficient method for aggregation of link metrics. The path for the delay minimization is determined utilizing a Minimum Delay Link Metric Aggregation Heuristic. Where path optimization is based Zo on maximum bandwidth, the link metric aggregation algorithm in accordance with the present invention is referred to as the Maximum Bandwidth Link Metric Aggregation Heuristic. Where path optimization is based on minimum delay, the link metric aggregation algorithm Zs in accordance with the present invention is referred to as the Minimum Bandwidth Link Mei:ric Aggregation Heuristic.
Thus, where link metrics are aggregated for bandwidth maximization utilizing the Maximum Bandwidth so Link Metric Aggregation Heuri~;tic, the following steps, shown in the flow chart of FIG. 11, numeral 1100, of are utilized: A) determining a maximum bandwidth path for each pair of exposed nodes irn the original subnetwork topology and representing each maximum bandwidth path 21~2~~1 by a virtual link whose link metric values equal respective metric values of the maximum bandwidth path and wherein a collection of all virtual links constitutes the full-mesh virtual topology (1102); B) determining, from the full-s mesh topology, a Maximum Bandwidth Spanning Tree using a Maximum Spanning Tree algorithm and bandwidth as link weights and distributing the bandwidth of the virtual links in the resulting spanning tree to the external nodes in the network (1104); and C) determining, from the full-mesh to topology, a Maximum Delay Spawning Tree using a Maximum Spanning Tree algorithm and clelay as link weights and distributing the delay of the virtual links in the resulting spanning tree to the external nodes in the network (1106).
is The delay associated with a maximum bandwidth path connecting a pair of exposed nodes is an upper bound on the minimum delay for any path connecting the same pair of exposed nodes because a path requiring less bandwidth than the maximum bandwidth may always have Zo the maximum bandwidth path as an alternative.
The bandwidth of any virtu<~I link not on the spanning tree is typically estimated by taking the minimum bandwidth among the virtual links along the unique path 2s that forms a loop with the virtual link, i.e., the bandwidth of the path. The maximum bandwidth of the virtual link not on the spanning tree cannot be larger than the estimate, or the tree will not tie a maximum bandwidth spanning tree. On the other hand, it cannot be smaller than so the estimate, for there is a contradiction if there exists an alternate path with a larger bandwidth than the maximum bandwidth of the virtual link. Hence, the bandwidth of the unique path is~ indeed the same as the bandwidth of the maximum bandwidth path or the ss corresponding virtual link in the full-mesh topology.
Where a Minimum Spanning Tree algorithm is utilized in the second step instead of a Maximum Spanning Tree algorithm, lower bound estimatE~s are obtained by taking the maximum bandwidth among the virtual links along the s unique path.
The delay of any virtual link not on the spanning tree is typically estimated by taking the minimum delay among the virtual links along the unique path that forms a loop io with the virtual link, i.e., not the delay of the path since delay is an additive link metric. The delay of the virtual link not on the spanning tree cannot be larger than the estimate, or the tree is not a maximum delay spanning tree. Hence, the estimate serves as an upper bound on the is delay of the maximum bandwidth path or the corresponding virtual link in the full-mesh virtual topology.
Thus, as shown in FIG. 12, numeral 1200, where link metrics are aggregated for delay minimization utilizing ao the Minimum Delay Link Metric flggregation Heuristic, the following steps are included: A) determining a minimum delay path for each pair of exposed nodes in the original subnetwork topology wherein eaclh minimum delay path is represented by a virtual link whose link metric values are zs equal to the respective metric values of the minimum delay path and a collection of all the virtual links forms a full-mesh virtual topology (1202); B) determining, from the full-mesh topology, a Minimum Bandwidth Spanning Tree using a Minimum Spanning free algorithm and so bandwidth as link weights and distributing the bandwidth of the virtual links in the resulting spanning tree to the external nodes in the network (12'04); and C) determining, from the full-mesh topology, a Maximum Delay Spanning Tree using a Maximum Spanning 'free algorithm and delay ss as link weights and distributing the delay of the virtual ms2~~1 links in the resulting spanning trE~e to the external nodes in the network (1206}.
The bandwidth associated with a minimum delay path s connecting a pair of exposed nodes is a lower bound on the maximum bandwidth for any path connecting the same pair of exposed nodes because a path that can tolerate a larger delay than the minimum delay always has the minimum delay path as an alternative.
to The bandwidth of any virtual link not on the spanning tree is typically estimated by taking the maximum bandwidth among the virtual links along the unique path that forms a loop with the virtual link, i.e., not the is bandwidth of the path. The maximum bandwidth of the virtual link not on the spanning tree cannot be smaller than the estimate, or the tree would not be a minimum bandwidth spanning tree. Hence, the estimate serves as a lower bound on the bandwidth o~f the minimum delay path Zo or the corresponding virtual link in the full-mesh topology.
The delay of any virtual link that is not on the spanning tree may be estimated by taking the minimum delay among the virtual links along the unique path that Zs forms a loop with the virtual link, i.e., not the delay of the path since delay is an additive link metric. The delay of the virtual link not on the spanning tree cannot be larger than the estimate, or the tree is not a maximum delay spanning tree. Hence, the estimate serves as an upper 3o bound on the delay of the minimum delay path or the corresponding virtual link in the full-mesh topology.
In both cases, the bounds on the link metrics of the virtual links may be somewhat flexible. When a particular 3s bound on a link metric is flexiible, one can additionally 2~~z4m advertise a specific virtual link to be used for replacing the estimate derived from the spanning tree. If additional metric values not associated with the spanning tree are advertised, it is important that they are indicated as such s in the advertisement, so that the spanning tree can be identified for generating metric estimates.
An example of the application of the invention using the Maximum Bandwidth Link Metric Aggregation Heuristic ~o is shown in FIGs. 4-9, numerals 400, 500, 600, 700, 800 and 900 respectively. FIG. 4, numeral 400, is a representation of exemplary bandwidth and delay metrics (402) in a subnetwork for utillizing the method of the present invention. FIG. 5, numeral 500, is a representation is of exemplary bandwidth and delay metrics (502) for a full-mesh topology derived from maximum bandwidth paths utilizing the method of the present invention. FIG. 6, numeral 600, is a representation of exemplary metrics (602) for a maximum bandwidth spanning tree in zo accordance with the method of the present invention. FIG.
7, numeral 700, is a representation of exemplary metrics (702) for a maximum delay spanning tree in accordance with the method of the present invention. FIG. 8, numeral 800, is a representation of exemplary metrics (802) Zs advertised for bandwidth and delay in accordance with the method of the present invention. FIG. 9, numeral 900, is a representation of exemplary estimates (902) for a full-mesh topology in accordance with the method of the present invention.
In the FIGs. 4-9, link metric; values associated with a link are represented by an orderE:d pair, (x, y), where x is the bandwidth value and y is the delay value, and wherever x or y is replaced by an "*", the corresponding link metric 3s value is not relevant in the context of the figure. Exposed ~162a~1 nodes A, B, C, and D are utilizE:d to construct a full-mesh topology. In this example, the bandwidth values are accurately recovered as expected, and upper bounds on the delay values are obtained. Where the delay estimate for s the virtual link AB differs considerably from the actual delay for the same virtual link, the delay value for that virtual link is typically additionally advertised together with an indication that it does not belong to the Maximum Delay Spanning Tree.
~o Thus, in one embodiment, chown in FIG. 13, numeral 1300, the method of the present invention for providing efficient aggregation of link metrics for a subnetwork in a communication network includes the steps of: A) selecting is a subset of nodes in a subnetwork of the communication network (1302} to be exposed nodes; B) reducing the original subnetwork topology to a full-mesh virtual topology (1304) connecting the exposed nodes, such that each virtual link connecting a pair of exposed nodes on the ao full-mesh virtual topology is deriived utilizing at least one of B1-B2: B1} determining a maximum bandwidth path in the original subnetwork topology and representing each maximum bandwidth path by <~ virtual link whose link metric values equal respective metric values of the Zs maximum bandwidth path, and wherein a collection of all virtual links constitutes the full-mesh topology; B2) determining a minimum delay path in the original subnetwork topology and representing each minimum delay path by a virtual link whose link metric values equal so respective metric values of the minimum delay path, and wherein a collection of all virltual links constitutes the full-mesh topology; C) where the full-mesh virtual topology is determined with 'the maximum bandwidth objective, C1 ) determining the spanning tree for encoding z~s2~9~
and advertising the bandwidth metrics (1306) by means of:
C1 a) determining, from the full-mesh topology, a Maximum Bandwidth Spanning Tree using a Maximum Spanning Tree algorithm and bandwidth as link weights; and C1 b) s distributing the bandwidth of the virtual links in the Maximum Bandwidth Spanning Tree to the external nodes in the network; C2) determinin~a the spanning tree for encoding and advertising the delay metrics by means of:
C2a) determiningi from the full-nnesh topology, a Maximum io Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights; and C2b) distributing the delay of the virtual links in the Maximum Delay Spanning Tree to the external nodes in the network; D) where the full-mesh virtual topology is determined with is the minimum delay objective, D1 ) determining the spanning tree for encoding and advertising the bandwidth metrics (1308) by means of: D1 a) determining, from the full-mesh topology, a Minimum Bandwidth Spanning Tree using a Minimum Spanning Tree algorithm and bandwidth as link 2o weights; and D1 b) distributing the bandwidth of the virtual links in the Minimum Bandwidi:h Spanning Tree to the external nodes in the network; D2) determining the spanning tree for encoding and advertising the delay metrics by means of: D2a) determining, from the full-as mesh topology, a Maximum Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights; D2b) distributing the bandwidth of the virtual links in the Maximum Delay Spainning Tree to the external nodes in the network.
FIG. 14, numeral 1400, is a flow chart of one embodiment of, in the step of assigning link metric values to the virtual links in accordance with the present invention, selecting a virtual link:, on the full-mesh virtual ...... 20 ~1.6~~~1 topology, whose metric value is to be estimated from the corresponding spanning tree, including the steps of: A) determining whether the virtual link is on the spanning tree (1402); B) where the virtual link is on the s spanning tree, a metric value of the virtual link on the spanning tree is taken to be an estimate for the metric value of the corresponding virtual link on the full-mesh topology (1404); C) where the virtual link is absent from the spanning tree, determining a unique path connecting io the exposed nodes attached at ends of the virtual link (1406), and determining whether the spanning tree is a maximum spanning tree (1408); C1 ) where the spanning tree is a maximum spanning tree, taking a minimum metric value among virtual links along a unque path to be an is estimate for a metric value of the virtual link on the full-mesh virtual topology (1410); and C2) where the spanning tree is other than a maximum spanning tree, i.e., a minimum spanning tree, taking a maximum metric value among virtual links along a unique path to be an estimate Zo for a metric value of the virtual link on the full-mesh virtual topology (1412).
Compared to the prior arir that divides nodes into master nodes and subordinate nodes such that the master as nodes maintain the network topology and carry out path computation on behalf of the subordinate nodes, the present invention provides the advantage of requiring little modification to existing routing protocols and architectures.
The present invention supports routing protocols that make use of link states, whereas in the prior art there are other methods that support routing protocols that make use of distance vectors. Aggregation of link states is zssz~~~_ more tractable than aggregation of distance vectors since the latter requires inter-subnetwork cooperation.
The present invention requires advertisement of s order M pieces of link state information when the number of exposed nodes is M. Unlike the approach that advertises a full-mesh representation of thE~ subnetwork topology, the invention does not suffer from link metric explosion. The estimated virtual topology proposed by the invention io adequately reflects asymmetry i~~n the original topology of the subnetwork.
While the prior art for routing to multiple destinations in a computer network tends to minimize the is cost of communicating with rnultiple destinations, the present invention minimizes the number of link metrics to be advertised and estimates link metrics that are not advertised, thereby providing efficient routing.
ao The unique method of the present invention provides for representing link metric values on a full-mesh topology by a spanning tree topology and for estimating link metric values for links that are not part of the spanning tree, a capability that is particularly useful in Zs link state aggregation for asynchronous transfer mode network implementations in an internetworking environment.
so I claim:
The prior art that makes no compromise in the s amount of link state information divides all nodes into two types: (i) master nodes with extended memory and computing capabilities, and (ii) subordinate nodes with limited memory and computing capabilities. A topology database is attached to each master node and is used by ~o the node for path computation. 'the subordinate nodes rely on the master nodes for topology maintenance and path computation.
In an extreme approach for link metric aggregation, ~s the entire subnetwork is repre:>ented by a pseudonode.
Such an approach can be traced back to the IS-IS, i.e., Intermediate System-to-Intermediate System, routing protocol. In one version of this approach, all internal link metrics of the subnetwork are reduced to a node metric for Zo the pseudonode by taking worst cases. For an additive link metric, a popular assignment of the corresponding node metric is the "diameter" or the length of the longest shortest path between any pair of nodes in the subnetwork.
For a non-additive link metric, a popular assignment of the Zs corresponding node metric is the bandwidth of the smallest maximum-bandwidth path between any pair of nodes in the subnetwork. Another version of the approach is to let the pseudonode to represent the "middle" of the subnetwork, and assign an appropriate metric, by taking 3o averages, to each of the links connecting the psuedonode and the exposed nodes. A variation of the above versions makes use of a diameter variance metric in addition to the diameter metric to reflect deviation from a symmetric topology.
216~~91 In the OSPF, i.e., the Open Shortest Path First, routing protocol, summary link states are used for inter-subnetwork routing. The summary links are essentially s virtual links to external nodes. From the point of view of a subnetwork, there are internal links that connect the internal nodes together, and there are summary links that emanate from the subnetwork to external nodes. OSPF is designed to be a link state routing protocol.
io In the prior art, a network is clustered into a number of levels, and each node has knowledge of only the shortest paths within its own lowest level cluster and a single shortest path to each supercluster. Path selection is is based on a distance-vector method, where a routing table indicating the best next hop for each intended destination is maintained by each node.
Also in the prior art that aggregates link states, the 20 original topology of the subnetwork is reduced to a smaller topology consisting of a subset of the nodes, namely exposed nodes, in the subnetwork. This method advertises a full-mesh virtual topology whose nodes are the exposed nodes and whose links represent paths in the original Zs topology. The paths are determined with respect to a given routing objective. The links in the virtual topology are known as virtual links. This approach is well suited to a hierarchical routing architecture.
3o In other prior art, a different problem is solved. A
heuristic that starts by computing a full-mesh shortest-path network consisting of a subset of the nodes in a given original network and then reducing the full-mesh network to a minimum spanning tree is used for routing to multiple destinations in a computer network, where the object is to minimize the total link weight on the multicast tree that connects the destination nodes together. There are two common spanning tree algorithms known in the prior art:
s maximum spanning tree algorithm and minimum spanning tree algorithm.
Although the approaches tlhat advertise the topology in terms of a pseudonode, i.e., variationsthe approach of in io IS-IS, offer the greatest reduction of advertised information, these approaches typically do not supply enough information for efficient routing.Moreover, the version that handles worst casE~s cannot reflect the link state information associated with an asymmetric is subnetwork topology. The version that takes averages cannot offer any performance guarantee.
The version that uses a diameter variance metric i s still under investigation. When a subset of the nodes in the subnetwork are exposed, all versions do apply.
not The approach that advertises the full-mesh representation of the subnetwork requires advertisement of order M2 pieces of link state information when the number of exposed nodes is M. For a subnetwork whose 2s nodes are not very densely connected, the amount of link state information to be advertised may well exceed that contained in the original topology of the subnetwork when M is sufficiently large. Such explosion of link state information contradicts the purpose of link metric so aggregation, i.e., efficiency in routing.
The summary link approach in OSPF does not capture enough link state information for efficient routing that is subject to quality of service constraints. Specifically, it 6 ~~s~~~1 does not offer the flexibility to pick and choose link resources that best meet the quality of service requirements for a connection.
s Efficient aggregation of link metrics should provide more efficient routing through the communication network.
Hence, there is a need for a method for efficient aggregation of link metrics such that the number of link metrics to be advertised is minimized without appreciably io compromising the information contained in the link metrics.
Brief Descriptions of the Drawings is FIG. 1 is a representation of a topology of a subnetwork for utilizing the method of the present invention.
ao FIG. 2 is a representation of a full-mesh topology of a subnetwork for utilizing the method of the present invention.
FIG. 3 is a representation o~f a spanning tree topology Zs of a subnetwork for utilizing the method of the present invention.
FIG. 4 is a representation of exemplary bandwidth and delay metrics in a subnetwork for utilizing the method 30 of the present invention.
FIG. 5 is a representation of exemplary bandwidth and delay metrics for a full-mesh topology derived from a... ~ ~~6~~91 maximum bandwidth paths utilizing the method of the present invention.
FIG. 6 is a representation of exemplary metrics for a s maximum bandwidth spanning tree in accordance with the method of the present invention.
FIG. 7 is a representation of exemplary metrics for a maximum delay spanning tree in accordance with the to method of the present invention.
FIG. 8 is a representation of exemplary metrics advertised for bandwidth and delay in accordance with the method of the present invention.
FIG. 9 is a representation ~of exemplary estimates for a full-mesh topology in accordance with the method of the present invention.
ao FIG. 10 is a flow chart of one embodiment of steps in accordance with the method of the present invention, which provides efficient aggregation of link metrics for a subnetwork in a communication network having a plurality of interconnected subnetworks that consist of nodes and 2s links.
FIG. 11 is a flow chart of one embodiment of steps for aggregation of link metrics fior bandwidth maximization utilizing the Maximum Bandwidth Link Metric so Aggregation Heuristic in accordance with the method of the present invention.
FIG. 12 is a flow chart of one embodiment of steps for aggregation of link metrics for delay minimization 2~.6~491 g utilizing the Minimum Delay Link Metric Aggregation Heuristic in accordance with the method of the present invention.
s FIG. 13 is a flow chart of a,n embodiment of steps in accordance with the method of the present invention.
FIG. 14 is a flow chart of one embodiment of, in the step of assigning link metric valn,~es to the virtual links in io accordance with the present invention, selecting a virtual link, on the full-mesh virtual topology, whose metric value is to be estimated from the corrE~sponding spanning tree.
~s Detailed Description of a Preferred Embodiment In a large internetworking environment consisting of many subnetworks, aggregation of link metrics would be advantageous, in particular for two reasons. First, the Zo amount of link state information must be compressed to reduce the topology update complexity which is proportional to the number of links. Second, the actual topology of a subnetwork may have to be hidden for security reasons. The method of the present invention is as applicable to link state routing protocols, and provides an efficient solution to the problem of aggregating link states, specifically link metrics, within a subnetwork where a predetermined subset of the nodes are exposed, i.e., to be retained in the process of aggregation. This 3o method requires significantlly less communication bandwidth for topology updates than the conventional method where every node in tlhe network is involved in maintaining and updating the entire topology of the network. The method of the present invention may also be applied recursively to a network with hierarchical structu res.
The purpose of link metric aggregation is to s represent the original topology in the subnetwork by a virtual topology that consists of only the exposed nodes and the smallest number of virtual links that can adequately capture the internal link metrics of the subnetwork for the purpose of path selection based on a io given routing objective. There may be one such virtual topology for each link metric. The virtual topologies do not have to be the same since only the link metric information derived from them is. of importance.
is For example, consider aggregating link metrics for a subnetwork with N connected nodes, M of which are exposed, or retained in the aggrecaation, where M < N, where M and N are positive integers. F=IG. 1, numeral 100, shows an exemplary subnetwork (102) whose exposed nodes (A, B, ao C, D) are shaded. For M exposed nodes, the largest number of virtual links that connect them together, since no parallel links are allowed, is M(M-1 )/2, i.e., order M2, and such a fully connected topology, FIG. 2, numeral 200, is known as a full-mesh topology (fully interconnected A-B-as C-D; 202). The smallest number of virtual links that connect all the exposed nodes i:ogether is (M-1 ), and any connected virtual topology with (M-1 ) virtual links belongs to the family of topologies known as "trees". Specifically, since all M exposed nodes are o~n the tree in this case, the so tree, shown in FIG. 3, numeral 300, is known as a "spanning tree". A tree topology consists of no loops. Also, there is a unique path connecting each pair of nodes on the spanning tree.
Where topology update complexity is not an issue, one would clearly want to advertise all the link metric information contained in the full-mesh virtual topology, thereby obtaining the best, i.e.,, most complete, network s topology without permitting parallel links. The representation of the original subnetwork topology by a full-mesh topology connecting a predetermined set of exposed nodes is referred to as a complete summarization of the subnetwork topology. Where selected, the io information contained in the complete summarization is compressed by means of encoding. The method of the present invention encodes they link metric information contained in the full-mesh topology, together with minimum compromise in the capability to recover it. The is order M2 pieces of link metric information are compressed to order M pieces. Given this encoded link metric information, one can obtain (M-1 ) pieces of the link metric information with complete accuracy and derive reasonable estimates of the remaining pieces in the form of lower ao bounds for bandwidth metrics and upper bounds for delay metrics. In some cases, the Estimates are so accurate that they are equal to the exact metric values. In this way the number of link metrics to be advertised is minimized without appreciably compromising the information as contained in the link metrics.
The aggregation and distrik~ution of link metrics for a subnetwork are typically executed by a representative node elected from the nodes in the subnetwork. This node 3o is referred to as the subnetwork leader. The link metric advertisement is to be receivE;d by nodes outside the subnetwork. These nodes are referred to as external nodes.
Upon receiving the advertisement, each external node is w :' 2~ g249't ~~
responsible for decoding or estimating the values of the aggregated link metrics of the subnetwork.
FIG. 10, numeral 1000, is a flow chart of one s embodiment of steps in accordance with the method of the present invention, which provides efficient aggregation of link metrics for a subnetwork in a communication network, comprising the steps of: A) selE:cting, by a predetermined subnetwork leader, a subset of the nodes in the subnetwork io to be exposed nodes (1002); B) for each link metric, generating, by the subnetwork leader, a spanning tree virtual topology connecting exposed nodes (1004); and C) assigning link metric values to~ the virtual links on the spanning tree and advertising the metric values in a is topology broadcast by the subnetwork leader (1006).
Where selected, one may add, after step (A) a step of reducing, by the subnetwork leader, an original subnetwork topology, stored in a topology database of the subnetwork leader, to a full-mesh virtual topology wherein each Zo virtual link connecting a pair of .exposed nodes on the full-mesh virtual topology is derived utilizing a path selection optimization (1008). The advertised link metric values in step (D) typically include a number of predetermined specific link metric values for replacing selected Zs estimates derived from each spanning tree.
Thus, the original topology of the communication subnetwork is reduced by the subnetwork leader to a full-mesh virtual topology. Connecting a set of M predetermined 3o exposed nodes, M a positive integer, such that the metric values on each virtual link on the full-mesh virtual topology are derived from a path selection optimization, e.g., a maximum bandwidth path or a minimum delay path.
Then, for each link metric, the subnetwork leader ~' ~2 :2162491 generates a spanning tree viri:ual topology from the full-mesh virtual topology for encoding the metric values on the full-mesh topology. ThE~ subnetwork leader then advertises the encoded link metric values, i.e., those s assigned to the virtual links on the spanning trees, to the external nodes via a topology broadcast. For each link metric, a full-mesh topology connecting the M exposed nodes is derived, by each extE~rnal node, by means of a decoding technique, set forth below, from the spanning io tree topology such that the link metric value associated with each virtual link on the derived full-mesh topology is an estimate of the link metric value associated with the corresponding virtual link in the original full-mesh representation of the subnetwork topology.
Although generally the link metric values associated with a link are substantially symmetrical in both directions of a link, the method of the present invention is also applicable where link metric values are not ao symmetrical. In the case of the generic link metrics bandwidth and delay, the link bandwidth may be taken to be the minimum bandwidth between the two directions, and the delay to be the maximum delay between the two directions. In the present invention, an estimate for Zs bandwidth is a lower bound, and' where it is a lower bound on the minimum bandwidth between the two directions of a link, the bandwidth estimation is the lower bound on bandwidth for both directions. I~n the present invention, an estimate for delay is an upper bound, and where it is an 3o upper bound on the maximum delay between the two directions, the estimate for delay is the upper bound on delay for both directions.
The present invention provides for selectively 3s utilizing, for example, one of two types of path selection ~3 zlsz4~~
optimization: bandwidth maximization and delay minimization. A path for the bandwidth maximization is determined utilizing a maximum bandwidth path algorithm.
A path for the delay minimization is determined utilizing a s minimum delay path algorithm. The maximum bandwidth path algorithm is a straightforward variation of the shortest path algorithm. The present invention also utilizes a graph algorithm , the Maximum Spanning Tree algorithm, and its equivalent version Minimum Spanning io Tree. The Maximum/Minimum Spanning Tree algorithm is used to determine a spanning tree with the maximum/minimum sum of link weights.
The present invention combines the selected path is optimizationand spanning algorii:hms in a unique manner to provide a new, efficient method for aggregation of link metrics. The path for the delay minimization is determined utilizing a Minimum Delay Link Metric Aggregation Heuristic. Where path optimization is based Zo on maximum bandwidth, the link metric aggregation algorithm in accordance with the present invention is referred to as the Maximum Bandwidth Link Metric Aggregation Heuristic. Where path optimization is based on minimum delay, the link metric aggregation algorithm Zs in accordance with the present invention is referred to as the Minimum Bandwidth Link Mei:ric Aggregation Heuristic.
Thus, where link metrics are aggregated for bandwidth maximization utilizing the Maximum Bandwidth so Link Metric Aggregation Heuri~;tic, the following steps, shown in the flow chart of FIG. 11, numeral 1100, of are utilized: A) determining a maximum bandwidth path for each pair of exposed nodes irn the original subnetwork topology and representing each maximum bandwidth path 21~2~~1 by a virtual link whose link metric values equal respective metric values of the maximum bandwidth path and wherein a collection of all virtual links constitutes the full-mesh virtual topology (1102); B) determining, from the full-s mesh topology, a Maximum Bandwidth Spanning Tree using a Maximum Spanning Tree algorithm and bandwidth as link weights and distributing the bandwidth of the virtual links in the resulting spanning tree to the external nodes in the network (1104); and C) determining, from the full-mesh to topology, a Maximum Delay Spawning Tree using a Maximum Spanning Tree algorithm and clelay as link weights and distributing the delay of the virtual links in the resulting spanning tree to the external nodes in the network (1106).
is The delay associated with a maximum bandwidth path connecting a pair of exposed nodes is an upper bound on the minimum delay for any path connecting the same pair of exposed nodes because a path requiring less bandwidth than the maximum bandwidth may always have Zo the maximum bandwidth path as an alternative.
The bandwidth of any virtu<~I link not on the spanning tree is typically estimated by taking the minimum bandwidth among the virtual links along the unique path 2s that forms a loop with the virtual link, i.e., the bandwidth of the path. The maximum bandwidth of the virtual link not on the spanning tree cannot be larger than the estimate, or the tree will not tie a maximum bandwidth spanning tree. On the other hand, it cannot be smaller than so the estimate, for there is a contradiction if there exists an alternate path with a larger bandwidth than the maximum bandwidth of the virtual link. Hence, the bandwidth of the unique path is~ indeed the same as the bandwidth of the maximum bandwidth path or the ss corresponding virtual link in the full-mesh topology.
Where a Minimum Spanning Tree algorithm is utilized in the second step instead of a Maximum Spanning Tree algorithm, lower bound estimatE~s are obtained by taking the maximum bandwidth among the virtual links along the s unique path.
The delay of any virtual link not on the spanning tree is typically estimated by taking the minimum delay among the virtual links along the unique path that forms a loop io with the virtual link, i.e., not the delay of the path since delay is an additive link metric. The delay of the virtual link not on the spanning tree cannot be larger than the estimate, or the tree is not a maximum delay spanning tree. Hence, the estimate serves as an upper bound on the is delay of the maximum bandwidth path or the corresponding virtual link in the full-mesh virtual topology.
Thus, as shown in FIG. 12, numeral 1200, where link metrics are aggregated for delay minimization utilizing ao the Minimum Delay Link Metric flggregation Heuristic, the following steps are included: A) determining a minimum delay path for each pair of exposed nodes in the original subnetwork topology wherein eaclh minimum delay path is represented by a virtual link whose link metric values are zs equal to the respective metric values of the minimum delay path and a collection of all the virtual links forms a full-mesh virtual topology (1202); B) determining, from the full-mesh topology, a Minimum Bandwidth Spanning Tree using a Minimum Spanning free algorithm and so bandwidth as link weights and distributing the bandwidth of the virtual links in the resulting spanning tree to the external nodes in the network (12'04); and C) determining, from the full-mesh topology, a Maximum Delay Spanning Tree using a Maximum Spanning 'free algorithm and delay ss as link weights and distributing the delay of the virtual ms2~~1 links in the resulting spanning trE~e to the external nodes in the network (1206}.
The bandwidth associated with a minimum delay path s connecting a pair of exposed nodes is a lower bound on the maximum bandwidth for any path connecting the same pair of exposed nodes because a path that can tolerate a larger delay than the minimum delay always has the minimum delay path as an alternative.
to The bandwidth of any virtual link not on the spanning tree is typically estimated by taking the maximum bandwidth among the virtual links along the unique path that forms a loop with the virtual link, i.e., not the is bandwidth of the path. The maximum bandwidth of the virtual link not on the spanning tree cannot be smaller than the estimate, or the tree would not be a minimum bandwidth spanning tree. Hence, the estimate serves as a lower bound on the bandwidth o~f the minimum delay path Zo or the corresponding virtual link in the full-mesh topology.
The delay of any virtual link that is not on the spanning tree may be estimated by taking the minimum delay among the virtual links along the unique path that Zs forms a loop with the virtual link, i.e., not the delay of the path since delay is an additive link metric. The delay of the virtual link not on the spanning tree cannot be larger than the estimate, or the tree is not a maximum delay spanning tree. Hence, the estimate serves as an upper 3o bound on the delay of the minimum delay path or the corresponding virtual link in the full-mesh topology.
In both cases, the bounds on the link metrics of the virtual links may be somewhat flexible. When a particular 3s bound on a link metric is flexiible, one can additionally 2~~z4m advertise a specific virtual link to be used for replacing the estimate derived from the spanning tree. If additional metric values not associated with the spanning tree are advertised, it is important that they are indicated as such s in the advertisement, so that the spanning tree can be identified for generating metric estimates.
An example of the application of the invention using the Maximum Bandwidth Link Metric Aggregation Heuristic ~o is shown in FIGs. 4-9, numerals 400, 500, 600, 700, 800 and 900 respectively. FIG. 4, numeral 400, is a representation of exemplary bandwidth and delay metrics (402) in a subnetwork for utillizing the method of the present invention. FIG. 5, numeral 500, is a representation is of exemplary bandwidth and delay metrics (502) for a full-mesh topology derived from maximum bandwidth paths utilizing the method of the present invention. FIG. 6, numeral 600, is a representation of exemplary metrics (602) for a maximum bandwidth spanning tree in zo accordance with the method of the present invention. FIG.
7, numeral 700, is a representation of exemplary metrics (702) for a maximum delay spanning tree in accordance with the method of the present invention. FIG. 8, numeral 800, is a representation of exemplary metrics (802) Zs advertised for bandwidth and delay in accordance with the method of the present invention. FIG. 9, numeral 900, is a representation of exemplary estimates (902) for a full-mesh topology in accordance with the method of the present invention.
In the FIGs. 4-9, link metric; values associated with a link are represented by an orderE:d pair, (x, y), where x is the bandwidth value and y is the delay value, and wherever x or y is replaced by an "*", the corresponding link metric 3s value is not relevant in the context of the figure. Exposed ~162a~1 nodes A, B, C, and D are utilizE:d to construct a full-mesh topology. In this example, the bandwidth values are accurately recovered as expected, and upper bounds on the delay values are obtained. Where the delay estimate for s the virtual link AB differs considerably from the actual delay for the same virtual link, the delay value for that virtual link is typically additionally advertised together with an indication that it does not belong to the Maximum Delay Spanning Tree.
~o Thus, in one embodiment, chown in FIG. 13, numeral 1300, the method of the present invention for providing efficient aggregation of link metrics for a subnetwork in a communication network includes the steps of: A) selecting is a subset of nodes in a subnetwork of the communication network (1302} to be exposed nodes; B) reducing the original subnetwork topology to a full-mesh virtual topology (1304) connecting the exposed nodes, such that each virtual link connecting a pair of exposed nodes on the ao full-mesh virtual topology is deriived utilizing at least one of B1-B2: B1} determining a maximum bandwidth path in the original subnetwork topology and representing each maximum bandwidth path by <~ virtual link whose link metric values equal respective metric values of the Zs maximum bandwidth path, and wherein a collection of all virtual links constitutes the full-mesh topology; B2) determining a minimum delay path in the original subnetwork topology and representing each minimum delay path by a virtual link whose link metric values equal so respective metric values of the minimum delay path, and wherein a collection of all virltual links constitutes the full-mesh topology; C) where the full-mesh virtual topology is determined with 'the maximum bandwidth objective, C1 ) determining the spanning tree for encoding z~s2~9~
and advertising the bandwidth metrics (1306) by means of:
C1 a) determining, from the full-mesh topology, a Maximum Bandwidth Spanning Tree using a Maximum Spanning Tree algorithm and bandwidth as link weights; and C1 b) s distributing the bandwidth of the virtual links in the Maximum Bandwidth Spanning Tree to the external nodes in the network; C2) determinin~a the spanning tree for encoding and advertising the delay metrics by means of:
C2a) determiningi from the full-nnesh topology, a Maximum io Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights; and C2b) distributing the delay of the virtual links in the Maximum Delay Spanning Tree to the external nodes in the network; D) where the full-mesh virtual topology is determined with is the minimum delay objective, D1 ) determining the spanning tree for encoding and advertising the bandwidth metrics (1308) by means of: D1 a) determining, from the full-mesh topology, a Minimum Bandwidth Spanning Tree using a Minimum Spanning Tree algorithm and bandwidth as link 2o weights; and D1 b) distributing the bandwidth of the virtual links in the Minimum Bandwidi:h Spanning Tree to the external nodes in the network; D2) determining the spanning tree for encoding and advertising the delay metrics by means of: D2a) determining, from the full-as mesh topology, a Maximum Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights; D2b) distributing the bandwidth of the virtual links in the Maximum Delay Spainning Tree to the external nodes in the network.
FIG. 14, numeral 1400, is a flow chart of one embodiment of, in the step of assigning link metric values to the virtual links in accordance with the present invention, selecting a virtual link:, on the full-mesh virtual ...... 20 ~1.6~~~1 topology, whose metric value is to be estimated from the corresponding spanning tree, including the steps of: A) determining whether the virtual link is on the spanning tree (1402); B) where the virtual link is on the s spanning tree, a metric value of the virtual link on the spanning tree is taken to be an estimate for the metric value of the corresponding virtual link on the full-mesh topology (1404); C) where the virtual link is absent from the spanning tree, determining a unique path connecting io the exposed nodes attached at ends of the virtual link (1406), and determining whether the spanning tree is a maximum spanning tree (1408); C1 ) where the spanning tree is a maximum spanning tree, taking a minimum metric value among virtual links along a unque path to be an is estimate for a metric value of the virtual link on the full-mesh virtual topology (1410); and C2) where the spanning tree is other than a maximum spanning tree, i.e., a minimum spanning tree, taking a maximum metric value among virtual links along a unique path to be an estimate Zo for a metric value of the virtual link on the full-mesh virtual topology (1412).
Compared to the prior arir that divides nodes into master nodes and subordinate nodes such that the master as nodes maintain the network topology and carry out path computation on behalf of the subordinate nodes, the present invention provides the advantage of requiring little modification to existing routing protocols and architectures.
The present invention supports routing protocols that make use of link states, whereas in the prior art there are other methods that support routing protocols that make use of distance vectors. Aggregation of link states is zssz~~~_ more tractable than aggregation of distance vectors since the latter requires inter-subnetwork cooperation.
The present invention requires advertisement of s order M pieces of link state information when the number of exposed nodes is M. Unlike the approach that advertises a full-mesh representation of thE~ subnetwork topology, the invention does not suffer from link metric explosion. The estimated virtual topology proposed by the invention io adequately reflects asymmetry i~~n the original topology of the subnetwork.
While the prior art for routing to multiple destinations in a computer network tends to minimize the is cost of communicating with rnultiple destinations, the present invention minimizes the number of link metrics to be advertised and estimates link metrics that are not advertised, thereby providing efficient routing.
ao The unique method of the present invention provides for representing link metric values on a full-mesh topology by a spanning tree topology and for estimating link metric values for links that are not part of the spanning tree, a capability that is particularly useful in Zs link state aggregation for asynchronous transfer mode network implementations in an internetworking environment.
so I claim:
Claims (18)
OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:
1. A method for providing efficient aggregation of link metrics for a subnetwork in a communication network with an original subnetwork topology, comprising the steps of:
1A) selecting, by a predetermined subnetwork leader, a subset of nodes in the subnetwork to be exposed nodes, 1B) mapping, by the predetermined subnetwork leader, the original subnetwork topology, stored in a topology database of the subnetwork leader, to a full-mesh virtual topology wherein each virtual link connecting a pair of exposed nodes on the full-mesh virtual topology is derived utilizing a predetermined path selection optimization, 1C) for each link metric that characterizes a preselected plurality of information transfer of links in the subnetwork, generating, by the subnetwork leader, a spanning tree virtual topology connecting the exposed nodes, 1D) determining link metric values to they virtual links on the spanning tree and advertising the metric values in a topology broadcast to nodes outside the subnetwork, by the subnetwork leader, and 1E) estimating for each link metric, by nodes outside the subnetwork, link metric values for all virtual links on the full-mesh virtual topology connecting the exposed nodes based on the link metric values for the virtual links on a corresponding spanning tree, wherein a link metric value associated with each virtual link on the full-mesh virtual topology is an estimate of a link metric value associated with a corresponding virtual link in the original subnetwork topology.
1A) selecting, by a predetermined subnetwork leader, a subset of nodes in the subnetwork to be exposed nodes, 1B) mapping, by the predetermined subnetwork leader, the original subnetwork topology, stored in a topology database of the subnetwork leader, to a full-mesh virtual topology wherein each virtual link connecting a pair of exposed nodes on the full-mesh virtual topology is derived utilizing a predetermined path selection optimization, 1C) for each link metric that characterizes a preselected plurality of information transfer of links in the subnetwork, generating, by the subnetwork leader, a spanning tree virtual topology connecting the exposed nodes, 1D) determining link metric values to they virtual links on the spanning tree and advertising the metric values in a topology broadcast to nodes outside the subnetwork, by the subnetwork leader, and 1E) estimating for each link metric, by nodes outside the subnetwork, link metric values for all virtual links on the full-mesh virtual topology connecting the exposed nodes based on the link metric values for the virtual links on a corresponding spanning tree, wherein a link metric value associated with each virtual link on the full-mesh virtual topology is an estimate of a link metric value associated with a corresponding virtual link in the original subnetwork topology.
2. The method of claim 1 wherein the path selection optimization is one of:
bandwidth maximization and delay minimization, where bandwidth and delay are link metrics.
bandwidth maximization and delay minimization, where bandwidth and delay are link metrics.
3. The method of claim 2 wherein:
4A) link metrics are aggregated for bandwidth maximization utilizing a Maximum Bandwidth Link Metric Aggregation Heuristic, and link metrics are aggregated for delay minimization utilizing a Minimum Delay Heuristic, and 4B) where link metrics are aggregated for maximizing bandwidth, a Maximum Spanning Tree algorithm is utilized to determine a spanning tree with a maximum sum of link weights, and where link metrics are aggregated for bandwidth minimization, a Minimum Spanning Tree algorithm is utilized to determine a spanning tree with a minimum sum of link weights.
4A) link metrics are aggregated for bandwidth maximization utilizing a Maximum Bandwidth Link Metric Aggregation Heuristic, and link metrics are aggregated for delay minimization utilizing a Minimum Delay Heuristic, and 4B) where link metrics are aggregated for maximizing bandwidth, a Maximum Spanning Tree algorithm is utilized to determine a spanning tree with a maximum sum of link weights, and where link metrics are aggregated for bandwidth minimization, a Minimum Spanning Tree algorithm is utilized to determine a spanning tree with a minimum sum of link weights.
4. The method of claim 3 wherein link metrics are aggregated for bandwidth maximization utilizing the Maximum Bandwidth Link Metric Aggregation Heuristic that includes the steps of:
5A) determining a maximum bandwidth path for each pair of exposed nodes in the original subnetwork topology and representing each maximum bandwidth path by a virtual link whose link metric values equal respective link metric values of the maximum bandwidth path and wherein a collection of all virtual links constitutes the full-mesh virtual topology, 5B) determining, from the full-mesh virtual topology, a Minimum Bandwidth Spanning Tree using a Minimum Spanning Tree algorithm and bandwidth as link weights and distributing the bandwidth of the virtual links in the resulting spanning tree to the nodes outside the subnetwork; and 5C) determining, from the full-mesh virtual topology, a Maximum Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights and distributing the delay of the virtual links in the resulting spanning tree to the nodes outside the subnetwork.
5A) determining a maximum bandwidth path for each pair of exposed nodes in the original subnetwork topology and representing each maximum bandwidth path by a virtual link whose link metric values equal respective link metric values of the maximum bandwidth path and wherein a collection of all virtual links constitutes the full-mesh virtual topology, 5B) determining, from the full-mesh virtual topology, a Minimum Bandwidth Spanning Tree using a Minimum Spanning Tree algorithm and bandwidth as link weights and distributing the bandwidth of the virtual links in the resulting spanning tree to the nodes outside the subnetwork; and 5C) determining, from the full-mesh virtual topology, a Maximum Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights and distributing the delay of the virtual links in the resulting spanning tree to the nodes outside the subnetwork.
5. The method of claim 4 wherein the bandwidth of a non-spanning tree virtual link on the Maximum Bandwidth Spanning Tree that is unselected by the Maximum Spanning Tree algorithm in step 5B is estimated by determining the minimum bandwidth among the virtual links along a unique path that forms a loop on the spanning tree with the virtual link.
6. The method of claim 4 wherein the delay of a non-spanning tree virtual link on the Maximum Delay Spanning Tree that is unselected by the Maximum Spanning Tree algorithm in step 5C is estimated by determining a minimum delay among the virtual links along a unique path that forms a loop on the spanning tree with the virtual link, wherein the minimum delay is an estimate of an upper bound on the delay of the maximum bandwidth path or a corresponding virtual link in the full-mesh virtual topology.
7. The method of claim 4 wherein link metrics are aggregated for delay minimization utilizing the Minimum Delay Link Metric Aggregation Heuristic that includes the steps of:
10A) determining a minimum delay path for each pair of exposed nodes in the original subnetwork topology and representing each minimum delay path by a virtual link whose link metric values are equal to the respective link metric values of the minimum delay path and wherein a collection of all the virtual links forms a full-mesh virtual topology, 10B) determining, from the full-mesh virtual topology, a Minimum Bandwidth Spanning Tree using a Minimum Spanning Tree algorithm and bandwidth as link weights and distributing the bandwidths of the virtual links in the resulting spanning tree to nodes outside the subnetwork, and 10C) determining, from the full-mesh virtual topology, a Maximum Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights and distributing the delay of the virtual links in the resulting spanning tree to the nodes outside the subnetwork.
10A) determining a minimum delay path for each pair of exposed nodes in the original subnetwork topology and representing each minimum delay path by a virtual link whose link metric values are equal to the respective link metric values of the minimum delay path and wherein a collection of all the virtual links forms a full-mesh virtual topology, 10B) determining, from the full-mesh virtual topology, a Minimum Bandwidth Spanning Tree using a Minimum Spanning Tree algorithm and bandwidth as link weights and distributing the bandwidths of the virtual links in the resulting spanning tree to nodes outside the subnetwork, and 10C) determining, from the full-mesh virtual topology, a Maximum Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights and distributing the delay of the virtual links in the resulting spanning tree to the nodes outside the subnetwork.
8. The method of claim 7 wherein the bandwidth of a non-spanning tree virtual link on the Minimum Bandwidth Spanning Tree that is unselected by the Minimum Spanning Tree algorithm in Step 10B is estimated by determining the maximum bandwidth along a unique path that forms a loop on the spanning tree with the virtual link, an estimate that serves as a lower bound on the bandwidth of the minimum delay path or a corresponding virtual link in the full-mesh virtual topology.
9. The method of claim 7 wherein the delay of a non-spanning tree virtual link on the Maximum Delay Spanning Tree that is unselected by the Maximum Spanning Tree algorithm in step 10C is estimated by determining the minimum delay among the virtual links along a unique path that forms a loop on the spanning tree with the virtual link, an estimate that serves as an upper bound on the delay of the minimum delay path or a corresponding virtual link in the full-mesh virtual topology.
10. The method of claim 1 wherein advertising the metric values provides advertised link metric values that include a number of predetermined specific link metric values for replacing selected estimates derived from each spanning tree in step 1E.
11. The method of claim 1 wherein the link metric values associated with each link are, as selected, one of: symmetrical and asymmetrical in both directions of the link.
12. The method of claim 11 wherein link metrics for the subnetwork include at least one of bandwidth and delay, and the following approximations are made:
17A) if bandwidth is a link metric, the bandwidth associated with a link is the minimum bandwidth between both directions of the link, and 17B) if delay is a link metric, the delay associated with a link is the maximum delay between both directions of the link.
17A) if bandwidth is a link metric, the bandwidth associated with a link is the minimum bandwidth between both directions of the link, and 17B) if delay is a link metric, the delay associated with a link is the maximum delay between both directions of the link.
13. The method of claim 12 wherein a bandwidth associated with a link is a minimum bandwidth between both directions of the link, and an estimate for a link metric bandwidth is a lower bound, the estimate for bandwidth is a lower bound on bandwidth for both directions.
14. The method of claim 12 wherein a delay associated with a link is the maximum delay between both directions of the link, and an estimate for a link metric delay is an upper bound, the estimate for delay is the upper bound on delay for both directions.
15. The method of claim 1 wherein advertising the link metric values in a topology update includes advertisement of order M pieces of link state information when the number of exposed nodes is M, where M is a positive integer that is less than or equal to the total number of nodes in the original subnetwork topology.
16. The method of claim 1 wherein determining link metric values for the virtual links includes, for a given link metric, selecting a virtual link, on the full-mesh virtual topology, whose metric value is to be estimated from the corresponding spanning tree utilizing the steps of:
A) determining whether the virtual link is on the spanning tree, B) where the virtual link is on the spanning tree, a metric value of the virtual link on the spanning tree is taken to be an estimate for the metric value of the corresponding virtual link on the full-mesh virtual topology, C) where the virtual link is absent from the spanning tree, determining a unique path connecting the exposed nodes attached at ends of the virtual link, and determining whether the spanning tree is a maximum spanning tree, C1) where the spanning tree is a maximum spanning tree, taking a minimum metric value among virtual links along a unique path to be an estimate for a metric value of the virtual link on the full-mesh virtual topology, and C2) where the spanning tree is other than a maximum spanning tree (i.e., a minimum spanning tree), taking a maximum metric value among virtual links along a unique path to be an estimate for a metric value of the virtual link on the full-mesh virtual topology.
A) determining whether the virtual link is on the spanning tree, B) where the virtual link is on the spanning tree, a metric value of the virtual link on the spanning tree is taken to be an estimate for the metric value of the corresponding virtual link on the full-mesh virtual topology, C) where the virtual link is absent from the spanning tree, determining a unique path connecting the exposed nodes attached at ends of the virtual link, and determining whether the spanning tree is a maximum spanning tree, C1) where the spanning tree is a maximum spanning tree, taking a minimum metric value among virtual links along a unique path to be an estimate for a metric value of the virtual link on the full-mesh virtual topology, and C2) where the spanning tree is other than a maximum spanning tree (i.e., a minimum spanning tree), taking a maximum metric value among virtual links along a unique path to be an estimate for a metric value of the virtual link on the full-mesh virtual topology.
17. A method for providing efficient aggregation of link metrics, including bandwidth and delay, for a subnetwork in a communication network having a plurality of interconnected subnetworks that consist of nodes and links, comprising the steps of:
A) selecting a subset of the nodes in a subnetwork of the communication network to be exposed nodes, B) mapping the original subnetwork topology to a full-mesh virtual topology such that each virtual link connecting a pair of exposed nodes on the full-mesh virtual topology is derived utilizing at least one of:
B1) determining a maximum bandwidth path connecting each pair of exposed nodes in the original subnetwork topology and representing each maximum bandwidth path by a virtual link whose link metric values equal respective metric values of the maximum bandwidth path and wherein a collection of all virtual links constitutes the full-mesh virtual topology, B2) determining a minimum delay path connecting each pair of exposed nodes in the original subnetwork topology and representing each minimum delay path by a virtual link whose link metric values equal respective metric values of the minimum delay path and wherein a collection of all victual links constitutes the full-mesh topology;
C) where the full-mesh virtual topology is determined with the maximum bandwidth objective, further utilizing the steps of:
C1) determining an encoding for bandwidth metrics associated with the full mesh virtual topology, and advertising encoded bandwidth metrics by means of:
C1a) determining, from the full-mesh virtual topology, a Maximum Bandwidth Spanning Tree using a Maximum Spanning Tree algorithm and bandwidth as link weights, and C1b) distributing the bandwidth of the virtual links in the Maximum Bandwidth Spanning Tree to nodes outside the subnetwork; and C2) determining an encoding for delay metrics associated with the full-mesh virtual topology, and advertising encoded delay metrics by means of:
C2a) determining, from the full-mesh topology, a Maximum Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights, and C2b) distributing the delay of the virtual links in the Maximum Delay Spanning Tree to nodes outside the subnetwork;
D) where the full-mesh virtual topology is determined with the minimum delay objective, further utilizing the steps of:
D1) determining an encoding for the bandwidth metrics associated with the full-mesh virtual topology, and advertising encoded bandwidth metrics by means of:
D1a) determining, from the full-mesh virtual topology, a Minimum Bandwidth Spanning Tree using a Minimum Spanning Tree algorithm and bandwidth as link weights, and D1b) distributing the bandwidths of the virtual links in the Minimum Bandwidth Spanning Tree to nodes outside the network; and D2) determining an encoding for delay metrics associated with the full-mesh virtual topology, and advertising encoded delay metrics by means of:
D2a) determining, from the full-mesh virtual topology, a Maximum Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights, and D2b) distributing the delay of the virtual links in the Maximum Delay Spanning Tree to nodes outside the subnetwork.
A) selecting a subset of the nodes in a subnetwork of the communication network to be exposed nodes, B) mapping the original subnetwork topology to a full-mesh virtual topology such that each virtual link connecting a pair of exposed nodes on the full-mesh virtual topology is derived utilizing at least one of:
B1) determining a maximum bandwidth path connecting each pair of exposed nodes in the original subnetwork topology and representing each maximum bandwidth path by a virtual link whose link metric values equal respective metric values of the maximum bandwidth path and wherein a collection of all virtual links constitutes the full-mesh virtual topology, B2) determining a minimum delay path connecting each pair of exposed nodes in the original subnetwork topology and representing each minimum delay path by a virtual link whose link metric values equal respective metric values of the minimum delay path and wherein a collection of all victual links constitutes the full-mesh topology;
C) where the full-mesh virtual topology is determined with the maximum bandwidth objective, further utilizing the steps of:
C1) determining an encoding for bandwidth metrics associated with the full mesh virtual topology, and advertising encoded bandwidth metrics by means of:
C1a) determining, from the full-mesh virtual topology, a Maximum Bandwidth Spanning Tree using a Maximum Spanning Tree algorithm and bandwidth as link weights, and C1b) distributing the bandwidth of the virtual links in the Maximum Bandwidth Spanning Tree to nodes outside the subnetwork; and C2) determining an encoding for delay metrics associated with the full-mesh virtual topology, and advertising encoded delay metrics by means of:
C2a) determining, from the full-mesh topology, a Maximum Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights, and C2b) distributing the delay of the virtual links in the Maximum Delay Spanning Tree to nodes outside the subnetwork;
D) where the full-mesh virtual topology is determined with the minimum delay objective, further utilizing the steps of:
D1) determining an encoding for the bandwidth metrics associated with the full-mesh virtual topology, and advertising encoded bandwidth metrics by means of:
D1a) determining, from the full-mesh virtual topology, a Minimum Bandwidth Spanning Tree using a Minimum Spanning Tree algorithm and bandwidth as link weights, and D1b) distributing the bandwidths of the virtual links in the Minimum Bandwidth Spanning Tree to nodes outside the network; and D2) determining an encoding for delay metrics associated with the full-mesh virtual topology, and advertising encoded delay metrics by means of:
D2a) determining, from the full-mesh virtual topology, a Maximum Delay Spanning Tree using a Maximum Spanning Tree algorithm and delay as link weights, and D2b) distributing the delay of the virtual links in the Maximum Delay Spanning Tree to nodes outside the subnetwork.
18. The method of claim 17 wherein advertising the link metric values in a topology update includes advertisement of order M pieces of link state information when the number of exposed nodes is M, where M is a positive integer that is less than or equal to the total number of nodes in the original subnetwork topology.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US08/239/178 | 1994-05-06 | ||
| US08/239,178 US5535195A (en) | 1994-05-06 | 1994-05-06 | Method for efficient aggregation of link metrics |
| PCT/US1995/001789 WO1995031058A1 (en) | 1994-05-06 | 1995-02-09 | Method for efficient aggregation of link metrics |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CA2162491C true CA2162491C (en) | 2000-02-22 |
Family
ID=22900962
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA002162491A Expired - Lifetime CA2162491C (en) | 1994-05-06 | 1995-02-09 | Method for efficient aggregation of link metrics |
Country Status (6)
| Country | Link |
|---|---|
| US (1) | US5535195A (en) |
| EP (1) | EP0711483B1 (en) |
| AU (1) | AU691260B2 (en) |
| CA (1) | CA2162491C (en) |
| DE (1) | DE69533535T2 (en) |
| WO (1) | WO1995031058A1 (en) |
Families Citing this family (124)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5793978A (en) | 1994-12-29 | 1998-08-11 | Cisco Technology, Inc. | System for routing packets by separating packets in to broadcast packets and non-broadcast packets and allocating a selected communication bandwidth to the broadcast packets |
| US5867666A (en) | 1994-12-29 | 1999-02-02 | Cisco Systems, Inc. | Virtual interfaces with dynamic binding |
| US6240451B1 (en) * | 1995-05-25 | 2001-05-29 | Punch Networks Corporation | Method and apparatus for automatically disseminating information over a network |
| US6097718A (en) | 1996-01-02 | 2000-08-01 | Cisco Technology, Inc. | Snapshot routing with route aging |
| US5600794A (en) * | 1995-08-04 | 1997-02-04 | Bay Networks, Inc. | Method and apparatus for managing exchange of metrics in a computer network by exchanging only metrics used by a node in the network |
| US6147996A (en) * | 1995-08-04 | 2000-11-14 | Cisco Technology, Inc. | Pipelined multiple issue packet switch |
| US6182224B1 (en) | 1995-09-29 | 2001-01-30 | Cisco Systems, Inc. | Enhanced network services using a subnetwork of communicating processors |
| US6091725A (en) | 1995-12-29 | 2000-07-18 | Cisco Systems, Inc. | Method for traffic management, traffic prioritization, access control, and packet forwarding in a datagram computer network |
| US6035105A (en) | 1996-01-02 | 2000-03-07 | Cisco Technology, Inc. | Multiple VLAN architecture system |
| US5652751A (en) * | 1996-03-26 | 1997-07-29 | Hazeltine Corporation | Architecture for mobile radio networks with dynamically changing topology using virtual subnets |
| US5831975A (en) * | 1996-04-04 | 1998-11-03 | Lucent Technologies Inc. | System and method for hierarchical multicast routing in ATM networks |
| US6308148B1 (en) | 1996-05-28 | 2001-10-23 | Cisco Technology, Inc. | Network flow data export |
| US6243667B1 (en) | 1996-05-28 | 2001-06-05 | Cisco Systems, Inc. | Network flow switching and flow data export |
| US6212182B1 (en) | 1996-06-27 | 2001-04-03 | Cisco Technology, Inc. | Combined unicast and multicast scheduling |
| US5802042A (en) * | 1996-06-28 | 1998-09-01 | Cisco Systems, Inc. | Autosensing LMI protocols in frame relay networks |
| US6434120B1 (en) | 1998-08-25 | 2002-08-13 | Cisco Technology, Inc. | Autosensing LMI protocols in frame relay networks |
| US6122283A (en) * | 1996-11-01 | 2000-09-19 | Motorola Inc. | Method for obtaining a lossless compressed aggregation of a communication network |
| US6304546B1 (en) | 1996-12-19 | 2001-10-16 | Cisco Technology, Inc. | End-to-end bidirectional keep-alive using virtual circuits |
| CA2196132C (en) * | 1997-01-28 | 2004-09-21 | Nicholas W. Dawes | Method for automatically aggregating objects |
| US6084858A (en) * | 1997-01-29 | 2000-07-04 | Cabletron Systems, Inc. | Distribution of communication load over multiple paths based upon link utilization |
| US6842780B1 (en) * | 1997-04-08 | 2005-01-11 | Swisscom Ag | Method of management in a circuit-switched communication network and device which can be used as a node in a circuit-switched communication network |
| JP3063721B2 (en) * | 1997-04-30 | 2000-07-12 | 日本電気株式会社 | Topology information exchange device and machine-readable recording medium recording program |
| US6356530B1 (en) | 1997-05-23 | 2002-03-12 | Cisco Technology, Inc. | Next hop selection in ATM networks |
| US6122272A (en) | 1997-05-23 | 2000-09-19 | Cisco Technology, Inc. | Call size feedback on PNNI operation |
| US6189043B1 (en) | 1997-06-09 | 2001-02-13 | At&T Corp | Dynamic cache replication in a internet environment through routers and servers utilizing a reverse tree generation |
| TW325618B (en) * | 1997-06-10 | 1998-01-21 | Inst Information Industry | Broadcasting processing method of wireless modem network group |
| US6078590A (en) | 1997-07-14 | 2000-06-20 | Cisco Technology, Inc. | Hierarchical routing knowledge for multicast packet routing |
| EP0892522A1 (en) * | 1997-07-15 | 1999-01-20 | Alcatel | Method and arrangement to aggregate parameter information in a hierarchically structured network |
| US6330599B1 (en) | 1997-08-05 | 2001-12-11 | Cisco Technology, Inc. | Virtual interfaces with dynamic binding |
| US6212183B1 (en) | 1997-08-22 | 2001-04-03 | Cisco Technology, Inc. | Multiple parallel packet routing lookup |
| US6512766B2 (en) | 1997-08-22 | 2003-01-28 | Cisco Systems, Inc. | Enhanced internet packet routing lookup |
| US6157641A (en) | 1997-08-22 | 2000-12-05 | Cisco Technology, Inc. | Multiprotocol packet recognition and switching |
| US6343072B1 (en) | 1997-10-01 | 2002-01-29 | Cisco Technology, Inc. | Single-chip architecture for shared-memory router |
| US6252855B1 (en) | 1997-12-22 | 2001-06-26 | Cisco Technology, Inc. | Method and apparatus for identifying a maximum frame size to maintain delay at or below an acceptable level |
| US6111877A (en) | 1997-12-31 | 2000-08-29 | Cisco Technology, Inc. | Load sharing across flows |
| WO1999040694A1 (en) * | 1998-02-05 | 1999-08-12 | Mitsubishi International Gmbh | Method for establishing a radio communication and telecommunication network |
| US6370121B1 (en) | 1998-06-29 | 2002-04-09 | Cisco Technology, Inc. | Method and system for shortcut trunking of LAN bridges |
| US6377577B1 (en) | 1998-06-30 | 2002-04-23 | Cisco Technology, Inc. | Access control list processing in hardware |
| US6308219B1 (en) | 1998-07-31 | 2001-10-23 | Cisco Technology, Inc. | Routing table lookup implemented using M-trie having nodes duplicated in multiple memory banks |
| US6182147B1 (en) | 1998-07-31 | 2001-01-30 | Cisco Technology, Inc. | Multicast group routing using unidirectional links |
| US6101115A (en) | 1998-08-07 | 2000-08-08 | Cisco Technology, Inc. | CAM match line precharge |
| US6389506B1 (en) | 1998-08-07 | 2002-05-14 | Cisco Technology, Inc. | Block mask ternary cam |
| US6377551B1 (en) * | 1998-08-17 | 2002-04-23 | Nortel Networks Limited | QoS based route determination method for communications networks |
| US20020085547A1 (en) * | 1998-09-17 | 2002-07-04 | Mcnamara Tod W. | High density network topology |
| US6246689B1 (en) | 1998-09-21 | 2001-06-12 | Lucent Technologies Inc. | Method and apparatus for efficient topology aggregation for networks with hierarchical structure |
| US6147971A (en) * | 1998-11-18 | 2000-11-14 | 3Com Corporation | Optimized routing method based on minimal hop count for use in PNNI based asynchronous transfer mode networks |
| US6614781B1 (en) | 1998-11-20 | 2003-09-02 | Level 3 Communications, Inc. | Voice over data telecommunications network architecture |
| US6493317B1 (en) * | 1998-12-18 | 2002-12-10 | Cisco Technology, Inc. | Traffic engineering technique for routing inter-class traffic in a computer network |
| US6771642B1 (en) | 1999-01-08 | 2004-08-03 | Cisco Technology, Inc. | Method and apparatus for scheduling packets in a packet switch |
| US6757791B1 (en) | 1999-03-30 | 2004-06-29 | Cisco Technology, Inc. | Method and apparatus for reordering packet data units in storage queues for reading and writing memory |
| US6760331B1 (en) | 1999-03-31 | 2004-07-06 | Cisco Technology, Inc. | Multicast routing with nearest queue first allocation and dynamic and static vector quantization |
| US6603772B1 (en) | 1999-03-31 | 2003-08-05 | Cisco Technology, Inc. | Multicast routing with multicast virtual output queues and shortest queue first allocation |
| US6600727B1 (en) | 1999-05-27 | 2003-07-29 | Cisco Technology, Inc. | Distributed network repeater system |
| US6631141B1 (en) | 1999-05-27 | 2003-10-07 | Ibm Corporation | Methods, systems and computer program products for selecting an aggregator interface |
| WO2000079806A1 (en) * | 1999-06-17 | 2000-12-28 | Marconi Communications Limited | Mesh networks |
| US6578086B1 (en) * | 1999-09-27 | 2003-06-10 | Nortel Networks Limited | Dynamically managing the topology of a data network |
| US6385174B1 (en) * | 1999-11-12 | 2002-05-07 | Itt Manufacturing Enterprises, Inc. | Method and apparatus for transmission of node link status messages throughout a network with reduced communication protocol overhead traffic |
| US6980537B1 (en) | 1999-11-12 | 2005-12-27 | Itt Manufacturing Enterprises, Inc. | Method and apparatus for communication network cluster formation and transmission of node link status messages with reduced protocol overhead traffic |
| US6349210B1 (en) | 1999-11-12 | 2002-02-19 | Itt Manufacturing Enterprises, Inc. | Method and apparatus for broadcasting messages in channel reservation communication systems |
| US6349091B1 (en) | 1999-11-12 | 2002-02-19 | Itt Manufacturing Enterprises, Inc. | Method and apparatus for controlling communication links between network nodes to reduce communication protocol overhead traffic |
| US6667976B1 (en) * | 1999-12-09 | 2003-12-23 | Lucent Technologies Inc. | Fuzzycast service in switches |
| US6650626B1 (en) * | 1999-12-10 | 2003-11-18 | Nortel Networks Limited | Fast path forwarding of link state advertisements using a minimum spanning tree |
| EP1135000A1 (en) * | 2000-03-17 | 2001-09-19 | Telefonaktiebolaget Lm Ericsson | Link aggregation |
| US6963927B1 (en) * | 2000-08-29 | 2005-11-08 | Lucent Technologies Inc. | Method and apparatus for computing the shortest path between nodes based on the bandwidth utilization link level |
| US8023421B2 (en) | 2002-07-25 | 2011-09-20 | Avaya Inc. | Method and apparatus for the assessment and optimization of network traffic |
| US7487237B2 (en) * | 2000-10-17 | 2009-02-03 | Avaya Technology Corp. | Load optimization |
| US7406539B2 (en) * | 2000-10-17 | 2008-07-29 | Avaya Technology Corp. | Method and apparatus for performance and cost optimization in an internetwork |
| US7539154B1 (en) * | 2000-10-17 | 2009-05-26 | Cisco Technology, Inc. | Method and apparatus to detect and break loop configuration |
| US7756032B2 (en) * | 2000-10-17 | 2010-07-13 | Avaya Inc. | Method and apparatus for communicating data within measurement traffic |
| US7080161B2 (en) * | 2000-10-17 | 2006-07-18 | Avaya Technology Corp. | Routing information exchange |
| US7363367B2 (en) * | 2000-10-17 | 2008-04-22 | Avaya Technology Corp. | Systems and methods for robust, real-time measurement of network performance |
| US7336613B2 (en) * | 2000-10-17 | 2008-02-26 | Avaya Technology Corp. | Method and apparatus for the assessment and optimization of network traffic |
| US7349994B2 (en) | 2000-10-17 | 2008-03-25 | Avaya Technology Corp. | Method and apparatus for coordinating routing parameters via a back-channel communication medium |
| US7720959B2 (en) * | 2000-10-17 | 2010-05-18 | Avaya Inc. | Method and apparatus for characterizing the quality of a network path |
| IL155355A0 (en) | 2000-10-17 | 2003-11-23 | Routescience Technologies Inc | Method and apparatus for performance and cost optimization in an internetwork |
| US7324647B1 (en) | 2000-10-23 | 2008-01-29 | Bbn Technologies Corp. | Quantum cryptographic key distribution networks with untrusted switches |
| US7023797B2 (en) * | 2000-12-30 | 2006-04-04 | Intel Corporation | Flexible aggregation of output links |
| US7158486B2 (en) * | 2001-03-12 | 2007-01-02 | Opcoast Llc | Method and system for fast computation of routes under multiple network states with communication continuation |
| US8086738B2 (en) * | 2007-05-24 | 2011-12-27 | Russell Fish | Distributed means of organizing an arbitrarily large number of computers |
| GB2379355B (en) * | 2001-08-31 | 2003-07-16 | Roke Manor Research | A method of deriving a metric for link in a network |
| US7254111B2 (en) * | 2001-10-05 | 2007-08-07 | Samsung Electronics Co., Ltd. | Routing coordination protocol for a massively parallel router architecture |
| US7152113B2 (en) * | 2001-10-19 | 2006-12-19 | Sun Microsystems, Inc. | Efficient system and method of node and link insertion for deadlock-free routing on arbitrary topologies |
| US7110389B2 (en) * | 2001-11-19 | 2006-09-19 | International Business Machines Corporation | Fanning route generation technique for multi-path networks |
| KR100411595B1 (en) * | 2001-12-24 | 2003-12-18 | 엘지전자 주식회사 | Method For Aggregating Link Information In The Private Network To Network Interface |
| US7203743B2 (en) * | 2001-12-28 | 2007-04-10 | Nortel Networks Limited | Hierarchical tree-based protection scheme for mesh networks |
| US7414985B1 (en) * | 2002-02-01 | 2008-08-19 | Ciena Corporation | Link aggregation |
| US7457416B1 (en) | 2002-07-17 | 2008-11-25 | Bbn Technologies Corp. | Key distribution center for quantum cryptographic key distribution networks |
| US20060222180A1 (en) * | 2002-10-15 | 2006-10-05 | Elliott Brig B | Chip-scale transmitter for quantum cryptography |
| US7627126B1 (en) | 2002-10-15 | 2009-12-01 | Bbn Technologies Corp. | Systems and methods for implementing path length control for quantum cryptographic systems |
| US7236597B2 (en) | 2002-12-20 | 2007-06-26 | Bbn Technologies Corp. | Key transport in quantum cryptographic networks |
| US7460670B1 (en) | 2002-12-20 | 2008-12-02 | Bbn Technologies Corp. | Systems and methods for managing quantum cryptographic networks |
| TWI254527B (en) * | 2003-02-19 | 2006-05-01 | Nec Corp | Network system, spanning tree configuration method and spanning tree configuration node |
| US7430295B1 (en) | 2003-03-21 | 2008-09-30 | Bbn Technologies Corp. | Simple untrusted network for quantum cryptography |
| US7706535B1 (en) | 2003-03-21 | 2010-04-27 | Bbn Technologies Corp. | Systems and methods for implementing routing protocols and algorithms for quantum cryptographic key transport |
| US7512242B2 (en) * | 2003-03-21 | 2009-03-31 | Bbn Technologies Corp. | Systems and methods for quantum cryptographic key transport |
| GB0306855D0 (en) * | 2003-03-25 | 2003-04-30 | Ideas Network Ltd | Data communication network |
| US7457303B2 (en) * | 2003-06-06 | 2008-11-25 | International Business Machines Corporation | One-bounce network |
| US7706282B2 (en) * | 2003-06-25 | 2010-04-27 | Leping Huang | Bluetooth personal area network routing protocol optimization using connectivity metric |
| US7580401B2 (en) * | 2003-10-22 | 2009-08-25 | Nortel Networks Limited | Method and apparatus for performing routing operations in a communications network |
| US20050129017A1 (en) * | 2003-12-11 | 2005-06-16 | Alcatel | Multicast flow accounting |
| US7333444B1 (en) * | 2003-12-17 | 2008-02-19 | Sun Microsystems, Inc. | Method and apparatus for creating a robust highly connected direct interconnection network |
| US7515716B1 (en) | 2004-02-26 | 2009-04-07 | Bbn Technologies Corp. | Systems and methods for reserving cryptographic key material |
| US7697693B1 (en) | 2004-03-09 | 2010-04-13 | Bbn Technologies Corp. | Quantum cryptography with multi-party randomness |
| US20060067337A1 (en) * | 2004-09-30 | 2006-03-30 | Netravali Arun N | Methods and devices for generating a hierarchical structure for the internet |
| US20060268691A1 (en) * | 2005-05-31 | 2006-11-30 | International Business Machines Corporation | Divide and conquer route generation technique for distributed selection of routes within a multi-path network |
| US7978719B2 (en) * | 2005-06-10 | 2011-07-12 | International Business Machines Corporation | Dynamically assigning endpoint identifiers to network interfaces of communications networks |
| US7978611B2 (en) * | 2005-09-06 | 2011-07-12 | At&T Intellectual Property I, L.P. | Systems and methods to determine network routes based on transmission medium length |
| US20070130455A1 (en) * | 2005-12-06 | 2007-06-07 | Elliott Brig B | Series encryption in a quantum cryptographic system |
| US20070133798A1 (en) * | 2005-12-14 | 2007-06-14 | Elliott Brig B | Quantum cryptography on a multi-drop optical network |
| US8082443B2 (en) * | 2006-01-09 | 2011-12-20 | Bbnt Solutions Llc. | Pedigrees for quantum cryptography |
| US7860106B2 (en) * | 2006-02-13 | 2010-12-28 | Wind River Systems, Inc. | System and method for routing table computation and analysis |
| US20070248092A1 (en) * | 2006-04-24 | 2007-10-25 | Motorola, Inc. | System and method for efficient traffic distribution in a network |
| US20080049701A1 (en) * | 2006-08-24 | 2008-02-28 | Sbc Knowledge Ventures, L.P. | System and method for calculating a bandwidth requirement between two elements in a communications network |
| US9049690B2 (en) * | 2006-12-27 | 2015-06-02 | Kyocera Corporation | Communication system, wireless communication terminal, communication method, wireless communication method, wireless communication apparatus and control method thereof |
| JP4894590B2 (en) * | 2007-03-30 | 2012-03-14 | ブラザー工業株式会社 | Network system, information processing apparatus, and information processing program |
| FR2915044B1 (en) * | 2007-04-16 | 2009-09-18 | France Telecom | METHOD FOR DETERMINING THE DYNAMICS OF A LOGIC NETWORK |
| US7930381B2 (en) * | 2007-12-04 | 2011-04-19 | International Business Machines Corporation | Efficient monitoring of heterogeneous applications |
| US9426228B2 (en) * | 2008-11-10 | 2016-08-23 | Nokia Technologies Oy | Cross layer service discovery for wireless networks |
| US8908526B2 (en) * | 2010-09-23 | 2014-12-09 | Intel Corporation | Controlled interconnection of networks using virtual nodes |
| US9949309B2 (en) * | 2011-02-19 | 2018-04-17 | Nomad Spectrum Limited | Methods, computer readable mediums, and apparatuses for providing communication to a mobile device using virtual connections |
| US8577380B2 (en) * | 2011-05-02 | 2013-11-05 | Telefonica, S.A. | Method for an automatic identification of urban dense areas from cell phones records |
| US9525638B2 (en) | 2013-10-15 | 2016-12-20 | Internap Corporation | Routing system for internet traffic |
| US10536366B1 (en) * | 2018-09-13 | 2020-01-14 | Charter Communication Operating, LLC | Methods and apparatus for controlling and making link bundle advertisements to support routing decisions |
| CN110286694B (en) * | 2019-08-05 | 2022-08-02 | 重庆邮电大学 | Multi-leader unmanned aerial vehicle formation cooperative control method |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4873517A (en) * | 1988-06-23 | 1989-10-10 | International Business Machines Corporation | Method for selecting least weight end node to end node route in a data communications network |
| JP2865675B2 (en) * | 1988-09-12 | 1999-03-08 | 株式会社日立製作所 | Communication network control method |
| US5079767A (en) * | 1988-09-27 | 1992-01-07 | Digital Equipment Corporation | Method of multicast message distribution |
| JP2531275B2 (en) * | 1989-09-29 | 1996-09-04 | 日本電気株式会社 | ATM cell transfer method |
| US5234592A (en) * | 1991-07-22 | 1993-08-10 | Wagner Spray Tech Corporation | Piston paint pump filter |
| US5274643A (en) * | 1992-12-11 | 1993-12-28 | Stratacom, Inc. | Method for optimizing a network having virtual circuit routing over virtual paths |
-
1994
- 1994-05-06 US US08/239,178 patent/US5535195A/en not_active Expired - Lifetime
-
1995
- 1995-02-09 CA CA002162491A patent/CA2162491C/en not_active Expired - Lifetime
- 1995-02-09 WO PCT/US1995/001789 patent/WO1995031058A1/en active IP Right Grant
- 1995-02-09 AU AU18426/95A patent/AU691260B2/en not_active Ceased
- 1995-02-09 DE DE69533535T patent/DE69533535T2/en not_active Expired - Lifetime
- 1995-02-09 EP EP95910242A patent/EP0711483B1/en not_active Expired - Lifetime
Also Published As
| Publication number | Publication date |
|---|---|
| AU691260B2 (en) | 1998-05-14 |
| EP0711483A1 (en) | 1996-05-15 |
| WO1995031058A1 (en) | 1995-11-16 |
| EP0711483B1 (en) | 2004-09-22 |
| DE69533535T2 (en) | 2005-01-27 |
| US5535195A (en) | 1996-07-09 |
| AU1842695A (en) | 1995-11-29 |
| DE69533535D1 (en) | 2004-10-28 |
| EP0711483A4 (en) | 1999-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA2162491C (en) | Method for efficient aggregation of link metrics | |
| US6567380B1 (en) | Technique for selective routing updates | |
| KR100218624B1 (en) | Routing bandwidth-reserved connections in information networks | |
| US7082473B2 (en) | System and method for optimizing open shortest path first aggregates and autonomous network domain incorporating the same | |
| US7796607B2 (en) | Scalable multiprotocol label switching based virtual private networks and methods to implement the same | |
| US7990979B2 (en) | Recursively partitioned static IP router tables | |
| US6614762B1 (en) | PNNI topology abstraction | |
| Garcia-Luna-Aceves et al. | Distributed, scalable routing based on vectors of link states | |
| CN103379032B (en) | The acquisition methods and device, sub-route computational entity of cross-domain end-to-end route | |
| US20070058607A1 (en) | Method and system for multi-layer network routing | |
| US20070242607A1 (en) | Method and system for controlling distribution of network topology information | |
| US20040085961A1 (en) | Method for advertising reachable address information in a network | |
| US7843837B2 (en) | Management of protocol information in PNNI hierarchical networks | |
| Baratz et al. | Establishing virtual circuits in large computer networks | |
| US6744734B1 (en) | Method for generating the optimal PNNI complex node representations for restrictive costs | |
| Bar-Noy et al. | Topology distribution cost vs. efficient routing in large networks | |
| US6731608B2 (en) | Complex node representations in PNNI systems | |
| CN114745227A (en) | Power service network slice time delay calculation method and device based on Flexe and SPN technologies | |
| Rekhter | IDRP protocol analysis: storage complexity | |
| Cidon et al. | Vicinity routing in large scale networks | |
| Braun et al. | A backup tree algorithm for multicast overlay networks | |
| Jeon | Topology aggregation method for multiple link parameters | |
| Tan et al. | A hierarchical routing model for large scale networks based on ant algorithm | |
| Ali et al. | Performance evaluation of hierarchical routing in ATM networks | |
| PANEL | QOS MANAGEMENT AND ROUTING |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| EEER | Examination request | ||
| MKEX | Expiry |
Effective date: 20150209 |