Intent driven root cause analysis
A fault model representation of a computer network is generated, wherein the computer network includes a set of connected computer network elements that was at least in part configured based on a specified declarative intent in forming the computer network. A symptom representation for the computer network is determined based on telemetry data of one or more elements of the set of connected computer network elements and a behavior specification repository identifying symptoms and their associated root causes. The fault model representation and the symptom representation are provided to a root cause analysis to determine one or more root causes of one or more detected symptoms of the computer network.
1. A method comprising:
generating a model representation of faults for a network based on an intent model for the network;
determining one or more declarative requirements for the network based on the model representation of faults for the network, wherein the one or more declarative requirements indicate a desired configuration of a plurality of network elements of the network;
determining one or more network elements from the plurality of network elements for telemetry monitoring based on the model representation of faults for the network and the one or more declarative requirements;
determining a symptom representation for the network based on telemetry data for the one or more network elements of the network; and
providing the model representation of the faults and the symptom representation to a root cause analysis engine to determine one or more root causes of one or more detected symptoms of the network.
2. The method of claim 1 , wherein generating the model representation of faults for the network includes representing at least a portion of the network as a graph representation of at least a portion of the set of network elements.
3. The method of claim 2 , wherein the graph representation comprises a plurality of nodes and at least one edge specifying a relationship between nodes.
4. The method of claim 1 , wherein the model representation of faults for the network models one or more faults associated with at least one of the following network elements: a switch, an interface, a link, or a protocol.
5. The method of claim 1 , wherein generating the model representation of faults for the network comprises deriving model representation of faults for the network at least in part from a representation that at least represents connections between the set of network elements.
6. The method of claim 1 , wherein at least a portion of the model representation of faults for the network is at least in part specified based on one or more network elements from a behavior specification repository.
7. The method of claim 1 , wherein at least a portion of the model representation of faults for the network is specified with a collection of elements that represents more than one distinct collection of underlying network elements.
8. The method of claim 1 , wherein at least a portion of the model representation of faults for the network is specified with elements that combine one or more relationships between network elements in a graph representation.
9. The method of claim 1 , wherein the telemetry data is collected by one or more processing agents upon a determination that one or more triggering patterns has occurred.
10. The method of claim 1 , wherein the telemetry data of the one or more network elements includes temporally aggregated data.
11. The method of claim 1 , wherein the telemetry data of the one or more network elements includes data of a first network element of the set of network elements that is combined with data of a second network element of the set of network elements.
12. The method of claim 1 , wherein the one or more network elements, for which the telemetry data is collected, are selected based at least in part on a service class determination for the one or more network elements.
13. The method of claim 1 , wherein determining the symptom representation for the network is further based on one or more root causes associated with a lack of connectivity between two connection endpoints.
14. The method of claim 1 , wherein a domain specific language specifies at least one of the following: the model representation of faults for the network, the symptom representation for the network, a behavior specification repository, or specified declarative intent for forming the network.
15. The method of claim 1 , wherein the network is at least in part configured using one or more processing agents.
16. The method of claim 15 , wherein one or more callback functions are used by the one or more processing agents upon a determination that one or more triggering patterns associated with the one or more callback functions has occurred.
17. The method of claim 1 , wherein the set of network elements is configured based on the intent model.
18. A system comprising:
a processor configured to:
generate a model representation of faults for a network based on an intent model for the network;
determine one or more declarative requirements for the network based on the model representation of faults for the network, wherein the one or more declarative requirements indicate a desired configuration of a plurality of network elements of the network;
determine one or more network elements from the plurality of network elements for telemetry monitoring based on the model representation of faults for the network and the one or more declarative requirements;
determine a symptom representation for the network based on telemetry data for the one or more network elements of the network; and
provide the model representation of faults for the network and the symptom representation to a root cause analysis engine to determine one or more root causes of one or more detected symptoms of the network; and
a memory coupled to the processor and configured to provide the processor with instructions.
19. A non-transitory computer-readable medium comprising instructions for causing a processor to:
generate a model representation of faults for a network based on an intent model for the network;
determine one or more declarative requirements for the network based on the model representation of faults for the network, wherein the one or more declarative requirements indicate a desired configuration of a plurality of network elements of the network;
determine one or more network elements from the plurality of network elements for telemetry monitoring based on the model representation of faults for the network and the one or more declarative requirements;
determine a symptom representation for the network based on telemetry data for the one or more network elements of the network; and
provide the model representation of faults for the network and the symptom representation to a root cause analysis engine to determine one or more root causes of one or more detected symptoms of the network.