U.S. patent application number 17/693523 was filed with the patent office on 2022-06-30 for topology partition detection.
The applicant listed for this patent is Arista Networks, Inc.. Invention is credited to Anthony Joseph Li.
Application Number | 20220210024 17/693523 |
Document ID | / |
Family ID | 1000006196621 |
Filed Date | 2022-06-30 |
United States Patent
Application |
20220210024 |
Kind Code |
A1 |
Li; Anthony Joseph |
June 30, 2022 |
TOPOLOGY PARTITION DETECTION
Abstract
Techniques described herein relate to a method for topology
partition detection. Network devices within a network topology
often share information to achieve desired network functionality.
However, when a network topology becomes partitioned such that not
all network devices are directly or indirectly connected to the
other network devices, the information sharing may fail. Techniques
described herein detect such network topology partitioning by
analyzing information that represents the network topology, and the
connections therein (e.g., a set of paths).
Inventors: |
Li; Anthony Joseph; (Los
Altos, CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Arista Networks, Inc. |
Santa Clara |
CA |
US |
|
|
Family ID: |
1000006196621 |
Appl. No.: |
17/693523 |
Filed: |
March 14, 2022 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
16739041 |
Jan 9, 2020 |
11296948 |
|
|
17693523 |
|
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
H04L 41/14 20130101;
H04L 43/0811 20130101; H04L 41/12 20130101 |
International
Class: |
H04L 41/12 20060101
H04L041/12; H04L 43/0811 20060101 H04L043/0811; H04L 41/14 20060101
H04L041/14 |
Claims
1. A method for topology partition detection, the method
comprising: making a first determination, by a network device of a
network, that a topology of the network has changed; obtaining,
based on the first determination, a set of paths from a current
topology for the network; assigning a first node of a first path of
the set of paths to a first partition; making a second
determination, using the first path, that a second node should be
connected to the first node; making a third determination, using
connectivity information of the network device, that the first node
and the second node are neighbors; and assigning, based on the
third determination, the second node to the first partition.
2. The method of claim 1, further comprising: making a fourth
determination, using the first path, that a third node in the first
path should be connected to the second node; making a fifth
determination, using the connectivity information, that the second
node and the third node are not neighbors; and assigning, based on
the fifth determination, the third node to a second partition;
making a sixth determination, using a second path of the set of
paths and the connectivity information that the second node and the
third node are neighbors and should be connected; and merging,
based on the sixth determination, the first partition and the
second partition into a third partition comprising the first node,
the second node, and the third node.
3. The method of claim 2, further comprising: making a seventh
determination that a fourth node of a third path of the set of
paths is not in the first partition; assigning, based on the
seventh determination, the fourth node to a fourth partition;
making an eighth determination that the third path does not include
any nodes that should be connected to the first node, the second
node, or the third node; making a ninth determination, based on the
eighth determination, that the current topology is partitioned into
the third partition and the fourth partition; and performing, based
on the ninth determination, a topology repair action set.
4. The method of claim 3, wherein performing the topology repair
action set comprises: performing an analysis on the connectivity
information to determine that the fourth node and the first node
are neighbors; and repairing, based on the analysis, the current
topology by updating the current technology to add a link between
the fourth node and the first node, wherein adding the link allows
for transmitting link state information between the first node and
the fourth node.
5. A method for topology partition detection, the method
comprising: making a first determination, by a network device in a
network, that a topology of the network has changed; obtaining, by
the network device and based on the first determination, a set of
paths from a current topology for the network; and performing, by
the network device, a partition detection action set using the set
of paths and a link state database of the network device, wherein
the partition detection action set determines whether the current
topology is partitioned.
6. The method of claim 5, wherein performing the partition
detection action set comprises: assigning a first node of a first
path of the set of paths to a first partition; making a second
determination, using the first path, that a second node should be
connected to the first node; making a third determination, using
the link state database, that the first node is a neighbor of the
second node; and assigning, based on the second determination and
the third determination, the second node to the first partition
with the first node.
7. The method of claim 5, wherein performing the partition
detection action set comprises: making a second determination,
using a first path of the set of paths, that a first node should be
connected to a second node; making a third determination, using the
link state database, that the first node is not a neighbor of the
second node; and assigning, based on the second determination and
the third determination, the second node in a second partition.
8. The method of claim 5, wherein performing the partition
detection action set comprises detecting that the current topology
is partitioned into a first partition and a second partition, and
the method further comprises: performing, in response to detecting
that the current topology is partitioned, a topology repair action
set.
9. The method of claim 8, wherein performing the topology repair
action set comprises: making a fourth determination, using the set
of paths, that a first node in the first partition and a second
node in the second partition are not connected in the current
topology; performing an analysis, using the link state database, to
determine that the first node is a neighbor of the second node; and
repairing, based on the analysis, the current topology by
transmitting link state information between the first node and the
second node.
10. The method of claim 5, wherein performing the partition
detection action set comprises detecting that the current topology
is partitioned into a first partition and a second partition, and
the method further comprises: making a second determination, using
the set of paths, that a first node of the first partition and a
second node of the second partition are connected in the current
topology making a third determination, using the link state
database that the first node and the second node are neighbors; and
merging, based on the second determination and the third
determination, the first partition and the second partition into a
third partition.
11. The method of claim 5, wherein performing the partition
detection action set determines that all nodes of the current
topology are in a single partition.
12. The method of claim 5 further comprising: performing, by the
network device in response to the first determination, an analysis
to determine routes to each of a plurality of other network devices
in the network.
13. The method of claim 5, wherein making the first determination
comprises receiving a link state update from a second device of the
network.
14. The method of claim 5, wherein making the first determination
comprises receiving an updated topology.
15. The method of claim 5, wherein the link state database
comprises link state information indicating whether nodes of the
current topology are neighbors.
16. A non-transitory computer readable medium comprising
instructions that, when executed by a processor, perform a method
for topology partition detection, the method comprising: making a
first determination, by a network device in a network, that a
topology of the network has changed; obtaining, by the network
device and based on the first determination, a set of paths from a
current topology for the network; and performing, by the network
device, a partition detection action set using the set of paths and
a link state database of the network device, wherein the partition
detection action set determines whether the current topology is
partitioned.
17. The non-transitory computer readable medium of claim 16,
wherein performing the partition detection action set comprises:
assigning a first node of a first path of the set of paths to a
first partition; making a second determination, using the first
path, that a second node should be connected to the first node;
making a third determination, using the link state database, that
the first node is a neighbor of the second node; and assigning,
based on the second determination and the third determination, the
second node to the first partition with the first node.
18. The non-transitory computer readable medium of claim 16,
wherein performing the partition detection action set comprises:
making a second determination, using a first path of the set of
paths, that a first node should be connected to a second node;
making a third determination, using the link state database, that
the first node is not a neighbor of the second node; and assigning,
based on the second determination and the third determination, the
second node in a second partition.
19. The non-transitory computer readable medium of claim 16,
wherein performing the partition detection action set comprises
detecting that the current topology is partitioned into a first
partition and a second partition, and the method further comprises:
performing, in response to detecting that the current topology is
partitioned, a topology repair action set.
20. The non-transitory computer readable medium of claim 19,
wherein performing the topology repair action set comprises: making
a fourth determination, using the set of paths, that a first node
in the first partition and a second node in the second partition
are not connected in the current topology; performing an analysis,
using the link state database, to determine that the first node is
a neighbor of the second node; and repairing, based on the
analysis, the current topology by transmitting link state
information between the first node and the second node.
Description
BACKGROUND
[0001] Networks of interconnected devices (e.g., computer networks)
often include any number of network devices. Such network devices
may be arranged in a network topology in order to provide
connectivity between devices within and outside of the network. A
network topology may be represented as a graph of interconnected
nodes. Within such network topologies, routing protocols are often
implemented that distribute certain information, at least in part,
by sharing the information (e.g., routing information) between
network devices (e.g., in packets) in the network topology. In
order for network devices in the network topology to share such
information, the information from each node (e.g., what neighbor
devices is the network device directly connected to, what does the
network device know about the topology, etc.) may be shared,
directly or indirectly, with the other nodes in the network
topology. However, conditions may exist in one or more network
devices (e.g., one or more network ports have failed, thereby
breaking link(s) between devices) that cause network devices to not
be operatively connected to each other. In such scenarios, one or
more network devices may no longer be operatively connected to
another device within the scheme being implemented for sharing
information among the network devices. This may cause the one or
more devices to be "partitioned" from the other network devices for
the purposes of information sharing, which may be referred to as a
partition of the graph that represents the network topology. Such a
partition may prevent a network from functioning or performing as
intended.
BRIEF DESCRIPTION OF DRAWINGS
[0002] FIG. 1 shows an example of a network topology in accordance
with one or more embodiments described herein.
[0003] FIG. 2 shows an example of an information sharing topology
in accordance with one or more embodiments described herein.
[0004] FIG. 3 shows a flowchart in accordance with one or more
embodiments described herein.
[0005] FIG. 4 shows a flowchart in accordance with one or more
embodiments described herein.
[0006] FIG. 5 shows a flowchart in accordance with one or more
embodiments described herein.
[0007] FIG. 6 shows an example in accordance with one or more
embodiments of described herein.
[0008] FIG. 7 shows an example in accordance with one or more
embodiments of described herein.
DETAILED DESCRIPTION
[0009] Specific embodiments will now be described with reference to
the accompanying figures.
[0010] In general, embodiments described herein relate to
techniques for detecting that a graph representing a network
topology, or subset therein, has become partitioned. In one or more
embodiments, a graph representing a network topology may be a graph
that includes all network devices in the network topology, and has
all the network devices operatively connected to one another (i.e.,
they are all operatively connected (i. e., directly or indirectly)
to each other network device in the network topology. For example,
the graph may represent a subset of the actual connections between
all the network devices, while still ensuring that the number of
connections represented allows the topology graph to meet certain
criteria. Examples of such criteria include, but are not limited
to: each network device is at least bi-connected to each other
network device (i.e., there are at least two paths between the
network devices); the topology graph has a degree (i.e., number of
connections from any network device) below a degree threshold; the
topology graph has a diameter (i.e., the longest path between any
two network devices) below a diameter threshold; etc. Such a
topology graph may, for example, represent a topology to be used
for sharing routing information among the network devices so that
each network device can maintain a synchronized copy of a link
state database for the network topology. As such, it may be
referred to as an information sharing topology.
[0011] Use of a topology represented by a topology graph that
includes only a subset of the actual connections in the network
topology may, for example, reduce the amount of information being
redundantly shared within the network topology, which may reduce
the load on the network and the devices therein (e.g., by reducing
link state packet flooding), improve network scalability, etc.
[0012] However, sharing information only on links represented as
connections between nodes in such a topology graph may lead to
scenarios where one or more network devices in the network topology
are not receiving the information they require for the network to
function as intended. Specifically, events may occur (e.g., link
failures, network device failures, network devices being removed
from service, links being added, network devices being added to the
network topology, etc.) that lead to an outcome where the
information being shared according to the topology graph no longer
causes all network devices in the network topology to receive the
information needed (e.g., routing information) for the network to
operate as intended. In such a scenario, one or more network
devices may be effectively partitioned from the other devices for
the purposes of receiving the needed information.
[0013] In one or more embodiments, embodiments described herein
allow for such partitions to be efficiently detected, so that
corrective actions may be taken to re-join any partitioned network
devices to the information sharing scheme represented, at least in
part, by the topology graph. Specifically, in one or more
embodiments, whenever a given network device in a network topology
determines that a network topology has changed in some way, the
network device initiates a procedure for determining if any one or
more network devices in the network topology are partitioned from
the other network devices for the purposes of sharing information,
such as routing information.
[0014] In one or more embodiments, the procedure for partition
detection includes obtaining, by a network device in the network
topology, a representation of the topology graph, which may be
represented, for example, as a set of paths. In one or more
embodiments, a path is a subset of the network devices in the
network topology, with the first network device in the path being
directly connected to the second network device, which is connected
to the third network device, and so on until the end of that
particular path is reached. Accordingly, in one or more
embodiments, a set of paths, when taken together, represent all of
the connections between network devices represented by the topology
graph that represents a subset (or all) of the connections in the
network topology. In one or more embodiments, the network devices
represented in a set of paths may be referred to as nodes.
[0015] In one or more embodiments, the network device then makes an
assumption that all network devices (i.e., nodes) in the network
topology are partitioned from each other. In one or more
embodiments, the network device then selects a path from the set of
paths, and checks a node database to see if the first node of the
path is already present therein. In one or more embodiments, a node
database is any data structure capable of including representations
of any kind that have been analyzed as part of the partition
detection described herein. In one or more embodiments, if the node
is present, then it has already been added to a partition, which a
logical grouping of one or more nodes in the node database based on
the node being connected (or not) to other nodes in the node
database. Such a logical partition of one or more nodes may be
included as additional information in the node database, or stored
as information in a separate data structure. In one or more
embodiments, any one or more data structures that include node
information and partition information may be collectively referred
to as a partition database. In one or more embodiments, if the node
is not present in the node database, a new partition is created,
and the node is added to that partition.
[0016] Next, in one or more embodiments, the next node in the path
is analyzed, using, for example, a link state database or a list of
neighbor nodes (collectively, connectivity information). In one or
more embodiments, the analysis determines whether the node is
connected to the previously analyzed node.
[0017] In one or more embodiments, if the node is connected to the
previous node, then the node database is checked to see if the node
is present therein. In one or more embodiments, if the node is
present, then a first determination is made as to the greatest
union that the node is present in, and a second determination is
made as to the greatest union the previous node is present in.
[0018] In one or more embodiments, a union is a logical joining of
two or more partitions. In one or more embodiments, a partition may
be added to a union with another partition based on at least one
node in a first partition being connected (per the connectivity
information) to at least one other in a second partition. In one or
more embodiments, a partition may be added to an existing union of
two or more partitions to make a new union of all the partitions.
In one or more embodiments, unions may be joined to form new unions
with all the nodes in all the partitions so joined being in the
union. In one or more embodiments, the union having the most nodes
(meaning all the nodes in the union are directly or indirectly
connected) may be referred to as the greatest union for a node that
has been analyzed and discovered to be connected to the previous
node in the path being analyzed and that was already present in the
node database at the time of the analysis. In one or more
embodiments, a union of partitions may also be referred to as a
partition, which includes all the nodes of the partitions that were
joined to create the union. In one or more embodiments, such
information may also be stored in the partition database.
[0019] In one or more embodiments, if the greatest union of the
node being analyzed is the same as the greatest union that the
previous node is in, no action is required, as the nodes are
already in the same partition (which may be a union of other
partitions, as discussed above). In one or more embodiments, the
two greatest unions are not the same, then a new union is created
that includes the two greatest unions (i. e., of the node being
analyzed and the previous node, which are connected to each other).
In one or more embodiments, the new union is a new partition that
includes all of the nodes in the two unions that were joined.
[0020] In one or more embodiments, if the node being analyzed,
which was found to be connected to the previous node, is not
present in the node database, then the node is added to the node
database and placed in the partition with the previous node.
[0021] In one or more embodiments, if the analysis of a node
determines that it is not connected to the previous node, then the
node database is consulted to determine if the node is already
present therein. In one or more embodiments, if the new node is
already present, then no further action is required, as the node is
already in a separate partition than the previous node. In one or
more embodiments, if the node is not present, then the node is
added to the node database and added to a new partition.
[0022] The above-described process for the node after the first
node is then repeated for each remaining node in the path, and the
above described process including the first node of a path is
repeated for all paths in the set of paths. In one or more
embodiments, if, at the end of the analysis of all the nodes in all
the paths in the set of paths leads to a partition database having
two or more partitions not joined together in a union, then the
information sharing topology represented by the set of paths is
considered partitioned. In one or more embodiments, if the
information sharing topology is partitioned, then one or more nodes
therein may fail to receive information updates necessary for
proper operation of the network topology. In one or more
embodiments, each network device in the network topology performs
the above-described analysis of partition detection
[0023] In one or more embodiments, once an information sharing
topology has been determined to be partitioned, corrective action,
if possible, may be taken to join the two or more partitions by
adding links to the information sharing topology represented by the
aforementioned topology graph. As an example, such corrective
action may include each network device in the network topology (all
of which performed the same analysis and came to the same
conclusion that the information sharing topology is partitioned)
consulting their common link state database and/or local neighbor
lists (connectivity information) to determine if they are, in fact,
directly connected to the partitioned network device even though
that connection is not represented in the set of paths that was
analyzed. In one or more embodiments, if a given network device is
connected to the partitioned network device, the connected network
device may add, at least temporarily, the partitioned network
device to the information sharing topology by adding the link
between the two devices, and begin to exchange information with the
now-previously partitioned network device.
[0024] In one or more embodiments, the above-described partition
detection allows an information sharing topology that has become
partitioned to be rapidly repaired. Such a method of partition
detection may also avoid the need to have a representation of the
entire information topology already existing in a data structure,
which would be required, for example, to perform a depth first
search or breadth first search of the information sharing topology.
Not requiring an existing data structure representing the entire
information sharing topology may conserve storage space on the
network devices in the network topology. Such a method of partition
detection may also allow the set of paths to each be analyzed only
once to determine if any the information sharing topology is
partitioned (e.g., as a result of a network topology change).
Efficiently analyzing the set of paths only once may reduce the
amount of time necessary to detect one or more partitioned network
devices relative to existing methods of partition detection.
[0025] FIG. 1 shows a network topology in accordance with one or
more embodiments described herein. The network topology shown in
FIG. 1 is only one example of a particular network topology. One
having ordinary skill in the art, and the benefit of this Detailed
Description, will appreciate that the techniques described herein
may apply to any network topology. Accordingly, embodiments
described herein should not be considered limited to the network
topology shown in FIG. 1.
[0026] As shown in FIG. 1, the network topology (100) is a
spine-leaf topology that includes a spine layer (102) and a leaf
layer (110). The spine layer (102) may include any number of spine
devices, such as spine device A (104), spine device B (106), and
spine device N (108). The leaf layer (110) may include any number
of leaf devices, such as leaf device A (112), leaf device B (114),
and leaf device N (116). Each of these components is described
below.
[0027] In one or more embodiments, a network topology (100) is an
arrangement of various elements of or within a network. In one or
more embodiments, a network includes a collection of one or more
network devices (e.g., spine devices (104, 106, 108), leaf devices
(112, 114, 116)) that facilitate network connectivity for one or
more operatively connected devices (e.g., computing devices, data
storage devices, etc. (not shown)). As used herein, a network may
be an entire network or any portion thereof (e.g., a logical area
of network devices within the network). A network may include a
datacenter network, a wide area network, a local area network, a
wireless network, a cellular phone network, or any other suitable
network that facilitates the exchange of information from one part
of the network to another. In one or more embodiments, the network
may be coupled with or overlap, at least in part, with the
Internet.
[0028] In one or more embodiments, the network topology (100) is a
bipartite network topology. In one or more embodiments, a bipartite
network topology is a network topology in which there are at least
two sets of network devices, with connections between devices of
one set and devices of another set, but no connections between
devices within a given set. Examples of such network topologies
include, but are not limited to, a folded three-stage Clos network
topology, and a spine-leaf topology. In one or more embodiments,
such network topologies may be referred to as dense network
topologies. In one or more embodiments, a dense network topology is
a network topology (100) in which the network devices are highly
interconnected, with, as an example, each device in a first set of
network devices being connected to each device in a second set of
network devices within the network topology. In the exemplary
embodiment shown in FIG. 1, the network topology (100) is a
spine-leaf topology. Although, FIG. 1 shows an example of a
spine-leaf topology, in one or more embodiments, the network
topology is not limited to being a spine-leaf topology (or the
particular example shown), or any other example of a dense network
topology set forth herein, and, as discussed above, may be any
network topology in which an information sharing topology exists
(e.g., for sharing routing information).
[0029] In one or more embodiments, when a network topology (100) is
arranged as a spine-leaf topology, the network topology (100)
includes a leaf layer (110) and a spine layer (102).
[0030] In one or more embodiments, a leaf layer (110) is a set of
any number of network devices (e.g., leaf devices (112, 114, 116))
that provide network connectivity to any number of attached
devices, such as computing devices (not shown). In one or more
embodiments, a computing device is any device or any set of devices
capable of electronically processing instructions and that
includes, at least, one or more processors, memory, input and
output device(s), and network connectivity via a leaf layer (110).
Examples of computing devices include, but are not limited to, a
server (e.g., a blade-server in a blade-server chassis, a rack
server in a rack, etc.), a virtual machine (VM), a desktop
computer, a mobile device (e.g., laptop computer, smart phone,
personal digital assistant, tablet computer, and/or any other
mobile computing device), and/or any other type of computing device
with the aforementioned requirements. Other examples of devices for
which a leaf layer (110) may provide network connectivity include,
but are not limited to, data storage devices (not shown), other
network devices (e.g., wireless routers) (not shown), media devices
(e.g., televisions) (not shown), etc.
[0031] In one or more embodiments, a spine layer (102) is a set of
any number of network devices (e.g., spine devices (104, 106, 108))
that provide network connectivity between network devices of a leaf
layer (110), and may also be operatively connected to other network
devices, such as edge routers (not shown), gateway routers (not
shown), network controllers (not shown), etc. In one or more
embodiments, each leaf device (112, 114, 116) within a leaf layer
(110) is operatively connected to each spine device (104, 106, 108)
within a spine layer (102), and each spine device (104, 106, 108)
within a spine layer is operatively connected to each leaf device
(112, 114, 116) within a leaf layer (110). Such connections may be
referred to as creating a dense mesh (e.g., a full mesh) between
the leaf layer (110) and the spine layer (102), thereby creating a
dense network topology (100).
[0032] In one or more embodiments, each leaf device (112, 114, 116)
and each spine device (104, 106, 108) is a network device. In one
or more embodiments, a network device may be a physical device that
includes and/or may operatively connected to persistent storage
(not shown), memory (e.g., random access memory (RAM)) (not shown),
one or more processor(s) (e.g., integrated circuits) (not shown),
and two or more physical network interfaces (which may also be
referred to as ports), which may provide connections (i. e., links)
to other devices (e.g., other network devices participating in a
routing protocol).
[0033] As shown in FIG. 1, examples of such interfaces include
ports S:1, S:2, and S:3 on each spine device (104, 106, 108) and
ports L:1, L:2, and L:3 on each leaf device (112, 114, 116).
Although not shown in FIG. 1, leaf devices (112, 114, 116) and/or
spine devices (104, 106, 108) may have any number of additional
ports.
[0034] In one or more embodiments, the network device also includes
any number of additional components (not shown), such as, for
example, network chips, field programmable gate arrays (FPGAs),
application specific integrated circuits (ASICs), indicator lights,
fans, etc.
[0035] In one or more embodiments, a network device includes
functionality to receive network traffic data units (e.g., frames,
packets, etc.) at any of the physical network interfaces (i.e.,
ports) of the network device and to process the network traffic
data units to determine whether to: (i) drop the network traffic
data unit; (ii) process the network traffic data unit; and/or (iii)
transmit the network traffic data unit, based on the processing,
from a physical port of the network device in accordance with one
or more embodiments.
[0036] As a non-limiting example, a network chip may be hardware
that receives network traffic data units at an ingress port, and
determines out of which egress port on the network device to
forward the network traffic data units such as, for example, media
access control (MAC) frames that may include Internet Protocol (IP)
packets.
[0037] In one or more embodiments, a network device (e.g., leaf
devices (112, 114, 116), spine devices (104, 106, 108)) may include
functionality to store (e.g., in persistent storage, in memory,
etc.), any number of data structures for facilitating operation of
at least some aspects of the network device. An example of such a
data structure is a link state database (not shown), which stores
routing protocol information such as the state of links (e.g., are
connections present) between devices in a network topology. Another
example of a data structure that may be stored on a network device
is a partition database (or set of databases) that is used to store
partition information during a partition detection analysis
(discussed further in the description of FIGS. 4-6, below).
[0038] In one or more embodiments, a link state database is a data
structure for storing link state information received from
neighboring network devices via the ports of a network device. In
one or more embodiments, network devices within a network topology
(100) (or a portion thereof (e.g., a logical area within the
topology)) share link state information using any of a variety of
types data units (e.g., packets, frames, messages, advertisements,
etc.). Such data units may be sent from one network device to
directly connected adjacent network devices, and may include
information related to the state of links between the sending
network device and other devices to which the sending network
device is operatively connected, the state of links between other
network devices in a network topology, etc.
[0039] When a data unit that includes link state information is
received at a network device from an adjacent network device, the
receiving network device may, in turn, propagate the data unit to
its own adjacent network devices. Such sharing of link state
information between network devices may occur within a network
topology implementing an interior gateway protocol (IGP). Examples
of an IGP include, but are not limited to, intermediate system to
intermediate system (IS-IS) and open shortest path first (OSPF),
each of which may be referred to as a type of IGP known as a
link-state routing protocol.
[0040] In one or more embodiments, when a link-state routing
protocol is implemented within a given network topology (or a
portion therein), each device participating in the link-state
routing topology receives, directly or indirectly, link state
information from other devices in the network topology, or logical
area within the network topology. In one or more embodiments, each
network device uses the received information to build a map of the
connectivity of network devices within the topology or logical
area. Information relating to the map may be stored in the link
state database.
[0041] A network device with a complete mapping of the connectivity
of network devices within a topology or logical area may then use
any scheme (e.g., the Dijkstra algorithm) to determine a best path
(e.g., shortest path, path with the least cost, etc.) from itself
to other network devices in the network topology or logical area,
and such information may be stored in the link state database
and/or used to program other portions of the network device (e.g.,
a routing information base) with information to be used, for
example, when processing a received network traffic data unit. In
one or more embodiments, the map of connectivity should be the same
on each network device in a topology or logical area, unless one or
more network devices have a change in information related to the
state of one or more links. Accordingly, once a mapping has been
determined, link state information may be sent from a given network
device only when a change to link state information of any link of
a network device occurs.
[0042] Link-state routing protocols often implement the sending of
link state information to adjacent devices in a process often
referred to as flooding, which is when the information is sent to
each adjacent connected network device, all of which then send to
each of their adjacent network devices, and so on, until all
network devices have received the link state information. However,
in a dense network topology (100), such flooding may lead to a
large quantity of redundant data units being transmitted to network
devices.
[0043] For example, in the spine-leaf network topology (100) shown
in FIG. 1, if a link state change is detected on leaf device A
(112), information related to the change may be flooded to each of
the three spine devices (104, 106, 108). Each of the spine devices
may then send the change to leaf device B (114) and leaf device N
(116), each of which may send the change back to the other spine
devices, etc. Such redundancy of receipt of link state information
when changes occur within a dense network topology may lead to
reduced performance of any given network device and/or the network
topology as a whole, because each redundant data item must be
processed in order to determine that it may ultimately be
discarded. As an example, the packet queues (not shown) of a
network device leading to a control plane (not shown) of the
network device are necessarily of finite size. Therefore, if the
flooding rate exceeds the update processing rate for enough time,
the control plan may be obligated to drop incoming updates. If such
lost updates are of significance, stabilization of the link state
database and the convergence of the network may be delayed.
[0044] In one or more embodiments, the possibility of reduced
performance is mitigated, at least in part, by only flooding link
state change information within a computed information sharing
topology, which uses only a subset of the links connecting network
devices within a network topology (100). Information sharing
topologies are discussed further in the description of FIG. 2,
below.
[0045] In one or more embodiments, a network device also includes
software and/or firmware stored in any network device storage (not
shown) and/or network device memory (not shown) (i.e.,
non-transitory computer readable mediums). Such software may
include instructions which, when executed by the one or more
processors (not shown) of the network device, cause the one or more
processors to perform operations in accordance with one or more
embodiments. The software instructions may be in the form of
computer readable program code to perform embodiments may be
stored, in whole or in part, temporarily or permanently, on a
non-transitory computer readable medium such as a CD, DVD, storage
device, a diskette, a tape, flash memory, physical memory, or any
other computer readable storage medium. Specifically, the software
instructions may correspond to computer readable program code that
when executed by a processor(s), is configured to perform
functionality related to embodiments. The functionality of a
network device is not limited to the aforementioned examples.
[0046] Examples of a network device include, but are not limited
to, a network switch, a router, a multilayer switch, a fibre
channel device, an InfiniBand.RTM. device, etc. A network device is
not limited to the aforementioned examples.
[0047] While FIG. 1 shows a configuration of components, other
configurations may be used without departing from the scope of
embodiments described herein. For example, the network topology may
be any network topology other than the spine-leaf topology shown in
FIG. 1. As another example, the quantity of network devices in the
spine layer may be less (or more) than the quantity of network
devices in the leaf layer. As another example, each leaf device and
each spine device may have any number of additional ports for
connecting to any number of other devices, or that may not be
connected to any other device at a given time. Accordingly,
embodiments disclosed herein should not be limited to the
configuration of components shown in FIG. 1.
[0048] FIG. 2 shows an information sharing topology (e.g., a
reduced information sharing topology) for sharing information
(e.g., routing information) within a network topology in accordance
with one or more embodiments described herein. The information
sharing topology shown in FIG. 2 is only one example of a
particular information sharing topology, and is described in the
context of one example of a scheme for information between network
devices using the information sharing topology. One having ordinary
skill in the art, and the benefit of this Detailed Description,
will appreciate that the techniques described herein may apply to
any network topology and/or any scheme for sharing information
within a network topology. Accordingly, embodiments described
herein should not be considered limited to the information sharing
topology shown in FIG. 2, or to the information sharing scheme
described below and in the description of FIG. 3.
[0049] As shown in FIG. 2, the information sharing topology (200)
includes a leaf layer (212) and a spine layer (202). The spine
layer (202) may include a leader (204), spine device B (206), spine
device C (208), and spine device D (210). The leaf layer may
include leaf device A (214), leaf device B (216), leaf device C
(218), and leaf device D (220). Each of these components is
described below.
[0050] In one or more embodiments, the leaf layer (212) is
substantially similar to the leaf layer (110) shown in FIG. 1 and
discussed above. Additionally, in one or more embodiments, the
spine layer (202) is substantially similar to the spine layer (102)
shown in FIG. 1 and discussed above.
[0051] In one or more embodiments, each of the leaf devices (214,
216, 218, 220) and each of the spine devices (204, 206, 208, 210)
are network devices and, as such, are substantially similar,
respectively, to the leaf devices (112, 114, 116) and the spine
devices (104, 106, 108) shown in FIG. 1 and discussed above.
However, for the sake of clarity, all of the connections between
each leaf device (214, 216, 218, 220) and each spine device (204,
206, 208, 210) are not shown in FIG. 2. Instead, only a subset of
the connections is shown. The subset may be referred to as an
information sharing topology (200).
[0052] In one or more embodiments, an information sharing topology
(200) is a subset of links within a network topology that includes
enough links such that there is at least one path between each
network device in the topology to each other network device in the
topology. In one or more embodiments, information (e.g., link state
updates) are only flooded on links included in the information
sharing topology (200), rather than to all links connected to
adjacent network devices, which may, for example, reduce the
redundancy of flooding within a network topology implementing a
link-state routing protocol. In one or more embodiments, an
information sharing topology (200) may include more than one path
between each device and every other device. For example, as shown
in FIG. 2, there are two paths between each network device and
every other network device, which may be referred to as a
bi-connected information sharing topology.
[0053] However, while the simple bi-connected topology shown in
FIG. 2 may be adequate as an information sharing topology (200) for
correctness and resiliency, such an information sharing topology
may not be optimal for convergence of link state databases on each
of the network devices within the topology, because the diameter
(i.e., the longest path between any two network devices in a
topology) of the information sharing topology may be half the
number of network devices in the topology, which may cause an
undesired delay in link state update propagation. Therefore, in one
or more embodiments, the actual information sharing topology may
have a bound on the diameter, which may be referred to as a
diameter threshold.
[0054] Additionally, although FIG. 2 shows an information sharing
topology (200) where each network device has two links used as part
of the information sharing topology, as more redundant paths
between nodes are added to an information sharing topology (200),
and when attempting to keep the diameter of the information sharing
topology below a diameter threshold, specific network devices may
end up with more links as part of the information sharing topology
than other devices. Having a high number of links as part of the
information sharing topology may cause reduced performance for the
network devices with a high number of links. Accordingly, when
computing an information sharing topology, in one or more
embodiments, the number of links within the information sharing
topology (i.e., the degree) of any given network device may be
bounded, which may be referred to as a degree threshold.
[0055] In one or more embodiments, at least one network device in
the information sharing topology is elected as a leader (204). In
one or more embodiments, a leader (204) is a network device elected
using any scheme for electing a leader within a network topology of
logical area of a network topology. In one or more embodiments, the
leader includes functionality to compute an information sharing
topology, encode the computed information sharing topology, and
distribute the encoded information sharing topology within a
network topology or logical area therein. Election of the leader
(204), as well as computation, encoding, and distribution of
information sharing topology (200), are discussed further in the
description of FIG. 3, below.
[0056] While FIG. 2 shows a configuration of components, other
configurations may be used without departing from the scope of
embodiments described herein. For example, as discussed above,
there may be more links in the subset of links that are the
information sharing topology, leading to more paths between network
devices (i.e., greater path redundancy). As another example, the
leader may be any device in the topology, rather than a spine
device as shown in FIG. 2. As another example, there may be
differing numbers of devices in the spine layer and the leaf layer.
As another example, there may be differing numbers of links on the
various network devices that are part of the information sharing
topology. Accordingly, embodiments disclosed herein should not be
limited to the configuration of components shown in FIG. 2.
[0057] FIG. 3 shows a flowchart describing a method for
distributing link state updates using an information sharing
topology in accordance with one or more embodiments described
herein. The method for creating and using an information sharing
topology shown in FIG. 3 is only one example of a particular scheme
for sharing information between network devices using the
information sharing topology. One having ordinary skill in the art,
and the benefit of this Detailed
[0058] Description, will appreciate that the techniques described
herein may apply to any network topology and/or any scheme for
sharing information within a network topology. Accordingly,
embodiments described herein should not be considered limited to
the information sharing topology creation and use shown in FIG.
3.
[0059] While the various steps in the flowchart shown in FIG. 3 are
presented and described sequentially, one of ordinary skill in the
relevant art, having the benefit of this Detailed Description, will
appreciate that some or all of the steps may be executed in
different orders, that some or all of the steps may be combined or
omitted, and/or that some or all of the steps may be executed in
parallel.
[0060] In Step 300, link state information is distributed by each
network device in a network topology in order to build a matching
link state database on each network device within the network
topology or logical area of a network topology. In one or more
embodiments, the initial link state database is built on each
network device using any link-state routing protocol scheme. For
example, both OSPF and IS-IS protocols cause the network devices
flood link state information from each network device to each
adjacent network device. In one or more embodiments, the link state
database includes information related to a mapping of the
connections between all network devices within the network topology
or logical area of the network topology.
[0061] In Step 302, once the link state database, including
connectivity information, has been built on each network device (i.
e., an initial convergence has occurred), a leader is elected from
among the network devices. In one or more embodiments, the leader
election may occur using any scheme for electing a single network
device from among the network devices in a network topology or
logical area of a network topology. Such schemes may include any
manner of indicating a leader priority, which may be any identifier
capable of indicating a relative priority level from among a group
of network devices related to which device should be a leader
network device. As an example, each network device may advertise a
leader priority, and the network device with the numerically
highest leader priority may be elected as the leader. In one or
more embodiments, if more than one network device advertises the
same highest leader priority, an election between such devices may
be made using any other identifying information of the network
devices. For example, from among the network devices advertising
the same highest leader priority, the elected leader may be the
network device with the numerically highest system identifier,
internet protocol address, router identifier, etc.
[0062] In Step 304, the leader computes the information sharing
topology. In one or more embodiments, computing the information
sharing topology includes computing a set of links from among all
the links of the network devices such that each network device has
at least two paths to each other network device. Any higher (or
lower) number of paths between any two network devices may be used
without departing from the scope of the embodiments described
herein. The determination of the number of paths between network
devices, and therefore the level of redundancy of paths, may
include a determination of a desired amount of path redundancy
versus an acceptable level of redundant distribution of link state
updates. Additionally, in one or more embodiments, computing the
information sharing topology includes ensuring that the diameter of
the information sharing topology does not exceed a diameter
threshold. Furthermore, in one or more embodiments, computing the
information sharing topology includes ensuring that any given
network device within the information sharing topology does not
exceed a degree threshold.
[0063] As an example, a computed information sharing topology may
be referred to as a minimal information sharing topology. In one or
more embodiments, a minimal information sharing topology is an
information sharing topology in which the topology is at least
bi-connected and each node has at least degree two, which may
guarantee that the information sharing topology has no single
points of failure. In a spine-leaf topology, a minimal information
sharing topology may cause every leaf device in the information
sharing topology to have a degree of two. As there are usually more
leaf devices than spine devices in a spine-leaf topology, the
degree of the spine devices may be higher than that of the leaf
devices. In one or more embodiments, the higher load on the spine
devices may be mitigated, at least in part, via an even
distribution of the load among the individual spine devices.
Accordingly, a minimal information sharing topology may scale well
with a diameter that is kept below a diameter threshold.
[0064] However, there may be alternate behaviors that can be
employed mitigate such risk. For example, if a leaf device sees
that its single link on the information sharing topology has
failed, the leaf device may compensate by performing a link state
database synchronization check with a different spine device (i.e.,
a spine device outside the known information sharing topology).
Similarly, if a leaf device determines that its connected spine
device within the information sharing topology has failed, it may
compensate by performing partition detection to determine if any
devices in the network topology have become partitioned from the
other network devices in the information sharing topology.
Partition detection is discussed further in the description of
FIGS. 4-6, below.
[0065] In Step 306, the information sharing topology is encoded. In
one or more embodiments, the information sharing topology is
encoded in any manner of encoding (i. e., representation of data)
that is able to represent the connections between the links of the
network devices that are included in the information sharing
topology. For example, if the information sharing topology was a
simple, bi-connected cycle such as that shown in FIG. 2, the
information sharing topology may be encoded as a list of nodes and
links in the information sharing topology, and/or as a set of paths
(discussed above in the description of FIG. 1). As another example,
the information sharing topology may be encoded using an adjacency
matrix. In one or more embodiments, an adjacency matrix is a matrix
in which rows represent one set of devices in a bipartite network
and the columns represent a second set of devices in the bipartite
network, with connections between devices that are included in the
information sharing topology being represented at the appropriate
intersection with a 1, and other intersections that are not within
the information sharing topology being represented as a 0.
[0066] In Step 308, the information sharing topology is distributed
to all network devices in the network topology or logical area of
the network topology using a current information sharing topology.
In one or more embodiments, the distribution of the information
sharing topology is performed using standard flooding (e.g.,
flooding as used in IS-IS or OSPF) so that all network devices
receive the information sharing topology. For example, the initial
distribution of the information sharing topology after the
information sharing topology is computed by the leader may be
performed using a standard information sharing topology. In one or
more embodiments, after the initial distribution of the information
sharing topology, subsequent distributions (e.g., after
re-computation of the information sharing topology) may be made
using the most current information sharing topology prior to the
re-computation, or the newly re-computed information sharing
topology. In one or more embodiments, each network device that
receives the information sharing topology stores the received
information sharing topology as a part of its link state database.
In one or more embodiments, any change in link state information
detected by any given network device is then transmitted from the
network device to adjacent network devices according to the
information sharing topology, and each adjacent network device that
receives such an update also transmits the update via the
information sharing topology.
[0067] In Step 310, a link state update is received by a network
device that is a part of the information sharing topology. In one
or more embodiments, the link state update is received from either
another network device on a link that is included in the
information sharing topology, or on a link that is outside the
information sharing topology (e.g., from a network device that is
not (yet) part of the information sharing topology).
[0068] In Step 312, the received link state update is propagated to
adjacent network devices according to the information sharing
topology. For example, if the information sharing topology dictates
that the network device that received the update in Step 310 should
transmit the update to two other network devices, then the link
state update is transmitted to only those two devices rather than
being flooded to all adjacent network devices. In one or more
embodiments, each of the adjacent devices also transmits the
received update according to the information sharing topology. In
one or more embodiments, the process continues to transmit the link
state update according to the information sharing topology until
all network devices of the information sharing topology, including
the leader, have received the link state update. In one or more
embodiments, although not shown in FIG. 3, if the received update
indicates a change in the network topology, the receiving network
device performs a partition detection. Partition detection is
discussed further in the descriptions of FIGS. 4-6, below
[0069] In Step 314, a determination is made, by the leader, whether
the information sharing topology should be re-computed. In one or
more embodiments, if the leader determines that the link state
update was initially sent from a network device that is already
within the information sharing topology, and the update does not
include a change that requires re-computation of the information
sharing topology, then the information sharing topology is not
re-computed, and the process may optionally end. Additionally, or
alternatively, as shown in FIG. 3, the process may optionally
return to Step 310 to await any additional updates that may be
received.
[0070] In one or more embodiments, when the leader determines that
the link state update that was received in Step 310 includes
contents that indicate that an updated information sharing topology
is required, then the process returns to Step 304, and the
information sharing topology is re-computed, re-encoded, and
re-distributed.
[0071] For example, a link status update may indicate that an
information sharing topology is no longer bi-connected, which
causes the leader to re-compute the information sharing topology to
satisfy the condition that the information sharing topology be at
least bi-connected. Other causes of information sharing topology
re-computation found in link state updates, such as information
related to degree threshold, diameter threshold, that all network
devices (e.g., nodes) of the network topology are no longer
covered, etc., may be used, alone or in combination, without
departing from the scope of embodiments described herein.
[0072] In one or more embodiments, if the update causing
re-computation of the information sharing topology was received
from a network device already included in the information sharing
topology, then the re-computed information sharing topology may be
sent using the current information sharing topology prior to the
re-computation or the new information sharing topology. In one or
more embodiments, if the update causing the leader to re-compute an
information sharing topology was received from a new network device
not previously included in the information sharing topology, then
the new network device from which the link state update was
initially received in Step 310 is included in the updated
information sharing topology, and the updated information sharing
topology is distributed throughout the network (e.g., including to
the new network device).
[0073] As an example of determining whether to re-compute an
information sharing topology, if a link is added to the topology,
the protocol being used will form a normal adjacency on the link
and update the appropriate link state advertisements for the
routers on either end of the link. These link state updates may
then be flooded on the information sharing topology. The leader,
upon receiving the updates, may choose to retain the existing
information sharing topology or may choose to modify the
information sharing topology. If the leader elects to change the
information sharing topology, it will update the information
sharing topology in the link state database and flood it using the
new information sharing topology.
[0074] FIGS. 4 and 5 show flowcharts describing a method for
partition detection in accordance with one or more embodiments
described herein. While the various steps in the flowchart shown in
FIGS. 4 and 5 are presented and described sequentially, one of
ordinary skill in the relevant art, having the benefit of this
Detailed Description, will appreciate that some or all of the steps
may be executed in different orders, that some or all of the steps
may be combined or omitted, and/or that some or all of the steps
may be executed in parallel.
[0075] In Step 400, a network device determines that a network
topology that the network device is within has changed. In one or
more embodiments, detecting a change in a network topology is
performed using any suitable scheme for determining that the
connections between network devices in a network topology have
changed in any way. As an example, a network device may receive a
link state update from a neighbor device, and, using the link state
database of the network device, determine that the link state
update indicates that the network topology has changed. As another
example, a network device may detect that a neighbor network device
is no longer connected (e.g., the neighbor network device
failed).
[0076] In Step 402, the network device obtains a set of paths that
represent an information sharing topology. In one or more
embodiments, as discussed above in the descriptions of FIG. 1 and
FIG. 2, an information sharing topology is a topology within the
network topology that represents how information (e.g., link state
information) is to be shared among the network devices of the
network topology, and may include only a subset of all links in the
network topology. In one or more embodiments, as discussed above in
the description of FIG. 1, the information sharing topology may be
represented as a set of paths, and may, for example, be stored as
part of the link state database of the network device.
[0077] In Step 406, the network device selects a path from the set
of paths. In one or more embodiments, any path in the set of paths
is suitable to be the first path or next path selected for
analysis. The network device may select the path using any scheme
for selecting a single instance from a group of such instances.
[0078] In Step 408, a determination is made whether the first node
in the path selected in Step 406 is present in a node database. As
described above, in one or more embodiments, a node database is any
data structure capable of including representations of any kind
that have been analyzed as part of the partition detection
described herein. In one or more embodiments, if the node is
present, then it has already been added to a partition, which a
logical grouping of one or more nodes in the node database based on
the node being connected (or not) to other nodes in the node
database. Such a logical partition of one or more nodes may be
included as additional information in the node database, and/or
stored as information in a separate data structure. In one or more
embodiments, any one or more data structures that include node
information (i.e., the node database) and partition information may
be collectively referred to as a partition database. In one or more
embodiments, if there the first node is already present in the node
database, then the method proceeds to Step 412. In one or more
embodiments, if the node is not present in the node database, then
the method proceeds to Step 410.
[0079] In Step 410, in one or more embodiments, if the node is not
present in the node database, the node is added to the node
database, a new partition is created, and the node is added to that
partition. In one or more embodiments, the network device, after
determining that the network topology is changed in Step 400, is
programmed to make an assumption that each node in the information
sharing topology has become partitioned from all other nodes in the
information sharing topology. In response to the assumption, the
network device, after determining that the first node of the
selected path is not already in a node database, creates a new
partition to include the node, and stores that partition in a
database. In one or more embodiments, a partition is a data
representation of one or more nodes in an information sharing
topology. In one or more embodiments, if there are more than one
node in a given partition, then the collection of nodes is
operatively connected, with each node having a direct connection to
at least one other node in the partition. In one or more
embodiments, after adding the node to the node database and to a
new partition, the method proceeds to Step 412.
[0080] In Step 412, a determination is made whether the path has
any next nodes.
[0081] In one or more embodiments, a path has a next node if there
is another node in the path that the path indicates should be a
neighbor of the previous node in the path (i.e., the next node is a
neighbor of the previous node). In one or more embodiments, if
there is a next node, the method proceeds to Step 414. In one or
more embodiments, if there is no next node in the path, the method
proceeds to Step 430.
[0082] In Step 414, a determination is made as to whether the next
node of the path found in Step 412 is connected to the previous
node in the path (i.e., it is a neighbor of the previous node). In
one or more embodiments, the determination is made using a
connectivity information from a data structure (e.g., a link state
database, a neighbor list, etc.) of the network device that
includes information relating to whether network devices are
neighbors to one another (i.e., are directly connected to one
another). In one or more embodiments, if the next node is a
neighbor of the previous node, the method proceeds to Step 416. In
one or more embodiments, if the next node is not a neighbor of the
previous node, the method proceeds to Step 424.
[0083] In Step 416, a determination is made whether the next node
is in the next node database. In one or more embodiments, if the
next node is in the node database, the method proceeds to Step 418.
In one or more embodiments, if the next node is not in the next
node database, the method proceeds to Step 424.
[0084] In Step 418, the greatest union of the next node, and the
greatest union of the previous node analyzed are computed. As
discussed above, in one or more embodiments, a union is a logical
joining of two or more partitions. In one or more embodiments, a
partition may be added to a union with another partition based on
at least one node in a first partition being connected (per the
connectivity information) to at least one other in a second
partition. In one or more embodiments, a partition may be added to
an existing union of two or more partitions no make a new union of
all the partitions being so joined. In one or more embodiments,
unions may be joined to form new unions with all the nodes in all
the partitions so joined being in the new union. In one or more
embodiments, the union having the most nodes (meaning all the nodes
in the union are directly or indirectly connected) may be referred
to as the greatest union for a node that has been analyzed and
discovered to be connected to the previous node in the path being
analyzed and that was already present in the node database at the
time of the analysis. In one or more embodiments, a union of
partitions may also be referred to as a partition, which includes
all the nodes of the partitions that were joined to create the
union. In one or more embodiments, such information may also be
stored in the partition database.
[0085] In Step 420, a determination is made whether the greatest
union of the next node and the greatest union of the previous node
are the same. In one or more embodiments, greatest unions are the
same if they include the same set of nodes. In one or more
embodiments, if the greatest union of the node being analyzed is
the same as the greatest union that the previous node is in, no
action is required, as the nodes are already in the same partition
(which may be a union of other partitions, as discussed above). In
one or more embodiments, if the greatest unions are the same, the
method returns to Step 412 to check for additional nodes. In one or
more embodiments, if the two greatest unions are not the same, then
the method proceeds to Step 422.
[0086] In Step 422, a new union is created that includes the two
greatest unions (i. e., of the node being analyzed and the previous
node, which are connected to each other). In one or more
embodiments, the new union is a new partition that includes all of
the nodes in the two unions that were joined. In one or more
embodiments, after the new union is created, the next node and the
previous node are in the same partition that is the new union.
Accordingly, the method then returns to Step 422 to check for
additional nodes.
[0087] In Step 424, after determining that the next node is not
connected to the previous node in Step 414, a determination is made
whether the next node is in the node database. In one or more
embodiments, if the next node is in the node database, then the
node is already in a partition, so no further action is necessary
with respect to the node, and the method returns to Step 412 to
check for additional nodes. In one or more embodiments, if the next
node is not in the node database, the method proceeds to Step
426.
[0088] In Step 426, in one or more embodiments, if the next node is
not present in the node database, the node is added to the node
database, a new partition is created, and the node is added to that
partition. In one or more embodiments, after adding the node to the
node database and to a new partition, the method returns to Step
412 to check for additional nodes.
[0089] In Step 428, after determining that the next node is
connected to the previous node in Step 414, but that the node is
not in the node database in Step 416, the node is added to the node
database, and added to the partition of the previous node. In one
or more embodiments, after adding the node to the node database and
to the partition of the previous node, the method returns to Step
412 to check for additional nodes.
[0090] In Step 430, once a determination is made that there are no
next nodes in a path being analyzed by the network device, a
determination is made whether the set of paths includes any
additional paths. In one or more embodiments, if there are
additional paths, the method returns to Step 406, another path is
selected and analyzed as described above. In one or more
embodiments, if there are no paths, the method ends.
[0091] One having ordinary skill in the art, and the benefit of
this Detailed
[0092] Disclosure, will appreciate that although FIG. 4 shows a
method to be performed in response to a single determination that a
network topology has changed, the method may be repeated each time
such a network topology change is detected. Additionally, one
having ordinary skill in the art, and the benefit of this Detailed
Disclosure, will appreciate that although FIG. 4 shows the method
as performed on one network device, the method may be performed on
all or any portion thereof of network devices in a network
topology.
[0093] Continuing with FIG. 5, in Step 500, a determination is made
as to whether there are any nodes that are partitioned from other
all other nodes. In one or more embodiments, the determination is
made by assessing the partition database to determine if, after
performing the partition detection discussed above in the
description of FIG. 4, there are two or more partitions that exist
that have not been joined in a union. For example, if the partition
detection found a partition A, a partition B, a partition C, and a
partition D, partitions A and B had been joined in a union to
create partition E, and partitions D and C had been joined to
create partition F, but partitions E and F were not joined, then
the information sharing topology is partitioned. In one or more
embodiments, if the information sharing topology is partitioned,
the method proceeds to Step 502. In one or more embodiments, if
there are no partitioned nodes, then the information sharing
topology is still functional (i. e., all nodes should be receiving
the information being shared among the nodes), and the method
ends.
[0094] In Step 502, a topology correction action set is performed.
In one or more embodiments, a topology correction action set is any
one or more actions that may lead to a repair of the information
sharing topology, such that all network devices that should be in
the information sharing topology are receiving the information
being shared. An example of such an action set is for each device
in the network topology that performed the partition detection to
consult its database of connectivity information (e.g., link state
database) to determine if it has a connection to one or more
partitioned devices and, if so, to add, at least temporarily, a
link to the information sharing topology that spans the partitions,
thereby causing the relevant information being shared to be
exchanged with the partitioned one or more nodes.
[0095] FIG. 6 shows an example in accordance with one or more
embodiments described herein. The following example is for
explanatory purposes only and not intended to limit the scope of
embodiments described herein. Additionally, while the example shows
certain aspects of embodiments described herein, all possible
aspects of such embodiments may not be illustrated in this
particular example.
[0096] Referring to FIG. 6, consider a scenario in which an area
leader of a network topology has previously distributed a set of
paths (600) that includes three paths that collectively represent
an information sharing topology within a network topology, and the
set of paths (600) has been received by a network device (not
shown) in the network topology (not shown) that includes nodes (not
shown) numbered 1-8. In this scenario, although each of nodes 1-8
will perform a partition detection, this example will be from the
perspective of the network device that is node 5 in the information
sharing topology represented by the set of paths.
[0097] Path A indicates that node 1 should be connected to node 2,
and node 2 should be connected to node 3. Path B indicates that
node 4 should be connected to node 5, node 5 should be connected to
node 6, and node 6 should be connected to node 7. Path C indicates
that node 2 should be connected to node 4, node 4 should be
connected to node 6, node 6 should be connected to node 8, and node
8 should be connected to node 1.
[0098] In this scenario, the following links become broken: the
link between 5 and 6, the link between 6 and 8, and the link
between 8 and 1. The network device receives at least one link
state update from a neighbor device, and uses the link state
database to determine that, indeed, a network topology change has
occurred. Therefore, in response to the determination, the network
device begins a partition detection process.
[0099] The network device then selects the Path A
(1.fwdarw.2.fwdarw.3). The network device begins with node 1 (first
node in path A), and assign it a partition A (not shown).
[0100] Next, the network device consults its link state database,
and finds that it shows that node 2 is connected to node 1, so node
2 is placed into partition A with node 1.
[0101] Next, the network device consults its link state database,
and finds that it shows that node 3 is connected to node 2, so node
3 is placed into partition A with node 1 and node 2.
[0102] Next, the network device determines that there are no more
nodes in path A, so the network device moves on to Path B
(4.fwdarw.5.fwdarw.6.fwdarw.7).
[0103] Next, the network device determines that node 4 is not in
partition A, so the network device creates partition B, and puts
node 4 in partition B.
[0104] Next, the network device consults its link state database,
and finds that it shows that node 5 is connected to node 4, so node
5 is placed into partition B with node 4.
[0105] Next, the network device consults its link state database,
and finds that it shows that node 6 is not connected to node 5, and
that node 6 is not in any existing partition, so the network device
creates partition C (not shown), and node 6 is placed into
partition C.
[0106] Next, the network device consults its link state database,
and finds that it shows that node 7 is connected to node 6, so node
7 is placed into partition C with node 6.
[0107] Next, the network device determines that there are no more
nodes in path B, so the network device moves on to Path C
(2.fwdarw.4.fwdarw.4.fwdarw.4.fwdarw.6.fwdarw.4.fwdarw.8.fwdarw.4.fwdarw.-
1).
[0108] Next, the network device determines that node 2 is already
in partition A, so no action is required
[0109] Next, the network device consults its link state database,
and finds that it shows that the link between node 4 and node 2 is
working (they are neighbors in the link state database). The
network device then consults the partition database to discover
that node 2 is in partition A, and node 4 is in partition B, so we
create partition D (not shown) to create a union between partition
A and partition B that includes nodes 1-5.
[0110] Next, the network device consults its link state database,
and finds that it shows that the link between node 6 and node 4 is
working (they are neighbors in the link state database). The
network device then consults the partition database to discover
that node 4 is in partition D, and node 6 is in partition C, so we
create partition E to create a union between partition D and
partition C that includes nodes 1-7.
[0111] Next, the network device consults its link state database,
and finds that it shows that node 8 is not connected to node 6, and
that node 8 is not in any existing partition, so the network device
creates partition F (not shown), and node 8 is placed into
partition F.
[0112] Next, the network device consults its link state database,
and finds that it shows that node 1 is not connected to node 8, but
that node 1 is already in partition A, so no action is taken.
[0113] A graphic representation of the results of the
above-described partition detection is shown in FIG. 7 as
partitioned information sharing topology (700)
[0114] Based on the above-described partition detection, the
network device that is node 5 has determined that node 8 is
partitioned from the other devices in the information sharing
topology. Accordingly, node 5 consults its link state database and
determines that although it is not shown as connected to
partitioned node in the set of paths, it is, in fact, physically a
neighbor of node 8. Therefore, node 5 adds the link to node 8 to
the information sharing topology and begins exchanging link state
updates with node 8, thereby repairing the partitioning of the
information sharing topology.
Example Embodiments
[0115] In general, in one aspect, embodiments described herein
relate to a method for topology partition detection. In one or more
embodiments, the method includes making a first determination, by a
network device of a network, that a topology of the network has
changed; obtaining, based on the first determination, a set of
paths from a current topology for the network; assigning a first
node of a first path of the set of paths to a first partition;
making a second determination, using the first path, that a second
node should be connected to the first node; making a third
determination, using connectivity information of the network
device, that the first node and the second node are neighbors; and
assigning, based on the third determination, the second node to the
first partition.
[0116] In general, in one aspect, embodiments described herein
relate to a method for topology partition detection. In one or more
embodiments, the method includes making a first determination, by a
network device in a network, that a topology of the network has
changed; obtaining, by the network device and based on the first
determination, a set of paths from a current topology for the
network; and performing, by the network device, a partition
detection action set using the set of paths and a link state
database of the network device, wherein the partition detection
action set determines whether the current topology is
partitioned.
[0117] In general, in one aspect, embodiments described herein
relate to a non-transitory computer readable medium including
instructions that, when executed by a processor, perform a method
for topology partition detection. In one or more embodiments, the
method includes making a first determination, by a network device
in a network, that a topology of the network has changed;
obtaining, by the network device and based on the first
determination, a set of paths from a current topology for the
network; and performing, by the network device, a partition
detection action set using the set of paths and a link state
database of the network device, wherein the partition detection
action set determines whether the current topology is
partitioned.
[0118] In the above description, numerous details are set forth as
examples of embodiments described herein. It will be understood by
those skilled in the art, and having the benefit of this Detailed
Description, that one or more embodiments of embodiments described
herein may be practiced without these specific details and that
numerous variations or modifications may be possible without
departing from the scope of the embodiments described herein.
Certain details known to those of ordinary skill in the art may be
omitted to avoid obscuring the description.
[0119] In the above description of the figures, any component
described with regard to a figure, in various embodiments described
herein, may be equivalent to one or more like-named components
described with regard to any other figure. For brevity,
descriptions of these components will not be repeated with regard
to each figure. Thus, each and every embodiment of the components
of each figure is incorporated by reference and assumed to be
optionally present within every other figure having one or more
like-named components. Additionally, in accordance with various
embodiments described herein, any description of the components of
a figure is to be interpreted as an optional embodiment, which may
be implemented in addition to, in conjunction with, or in place of
the embodiments described with regard to a corresponding like-named
component in any other figure.
[0120] Throughout the application, ordinal numbers (e.g., first,
second, third, etc.) may be used as an adjective for an element
(i.e., any noun in the application). The use of ordinal numbers is
not to imply or create any particular ordering of the elements nor
to limit any element to being only a single element unless
expressly disclosed, such as by the use of the terms "before",
"after", "single", and other such terminology. Rather, the use of
ordinal numbers is to distinguish between the elements. By way of
an example, a first element is distinct from a second element, and
the first element may encompass more than one element and succeed
(or precede) the second element in an ordering of elements.
[0121] As used herein, the phrase operatively connected, or
operative connection, means that there exists between
elements/components/devices a direct or indirect connection that
allows the elements to interact with one another in some way. For
example, the phrase `operatively connected` may refer to any direct
(e.g., wired directly between two devices or components) or
indirect (e.g., wired and/or wireless connections between any
number of devices or components connecting the operatively
connected devices) connection. Thus, any path through which
information may travel may be considered an operative
connection.
[0122] While a limited number of embodiments have been described
herein, those skilled in the art, having benefit of this
disclosure, will appreciate that other embodiments can be devised
which do not depart from the scope of the embodiments described
herein. Accordingly, the scope of embodiments described herein
should be limited only by the attached claims.
* * * * *