Forecasting network KPIs
In one embodiment, a service receives input data from networking entities in a network. The input data comprises synchronous time series data, asynchronous event data, and an entity graph that that indicates relationships between the networking entities in the network. The service clusters the networking entities by type in a plurality of networking entity clusters. The service selects, based on a combination of the received input data, machine learning model data features. The service trains, using the selected machine learning model data features, a machine learning model to forecast a key performance indicator (KPI) for a particular one of the networking entity clusters.
1. A method comprising:
receiving, at a service, input data from networking entities in a network, wherein the input data comprises synchronous time series data, asynchronous event data, and an entity graph that that indicates relationships between the networking entities in the network;
for each key performance indicator (KPI) to be forecasted among a plurality of KPIs to be forecasted, clustering, by the service, the networking entities by type in a plurality of networking entity clusters based on characteristics of each of the networking entities, such that each KPI to be forecasted is assigned to a different networking entity cluster of the plurality of networking entity clusters and each of the networking entities is capable of being included in more than one of the plurality of networking entity clusters;
selecting, by the service and based on a combination of the received input data, machine learning model data features based on a particular one of the KPIs to be forecasted being assigned to a particular one of the network entity clusters;
training, by the service and using the selected machine learning model data features, a machine learning model to forecast the particular KPI to be forecasted for the particular networking entity cluster; and
causing, by the service, the trained machine learning model to be used for forecasting the particular KPI.
2. The method as in claim 1 , wherein the networking entities comprise at least one of: a router, a switch, a wireless access point, or an access point controller.
3. The method as in claim 1 , wherein the causing of the trained machine learning model to be used for forecasting the particular KPI comprises:
deploying, by the service, the trained machine learning model to one or more of the networking entities in the particular one of the networking entity clusters.
4. The method as in claim 1 , wherein the causing of the trained machine learning model to be used for forecasting the particular KPI comprises:
receiving, at the service, a KPI forecast request from one of the networking entities in the particular one of the networking entity clusters;
using, by the service and in response to receiving the KPI forecast request, the trained machine learning model to forecast the particular KPI; and
providing, by the service, the forecast KPI to the networking entity in the particular one of the networking entity clusters that sent the KPI forecast request.
5. The method as in claim 1 , wherein the particular KPI is indicative of at least one of: a processor load, a memory load, or a traffic load.
6. The method as in claim 1 , wherein selecting, by the service and based on the combination of the received input data, the machine learning model data features comprises:
using the entity graph to select a subset of the networking entities; and
selecting the combination of the received input data from among the subset of the networking entities.
7. The method as in claim 1 , further comprising:
using the forecast KPI to predict a tunnel failure in the network.
8. The method as in claim 1 , wherein the network is a wireless network.
9. An apparatus, comprising:
one or more network interfaces;
a processor coupled to the network interfaces and configured to execute one or more processes; and
a memory configured to store a process executable by the processor, the process when executed configured to:
receive input data from networking entities in a network, wherein the input data comprises synchronous time series data, asynchronous event data, and an entity graph that that indicates relationships between the networking entities in the network;
for each key performance indicator (KPI) to be forecasted among a plurality of KPIs to be forecasted, cluster the networking entities by type in a plurality of networking entity clusters based on characteristics of each of the networking entities, such that each KPI to be forecasted is assigned to a different networking entity cluster of the plurality of networking entity clusters and each of the networking entities is capable of being included in more than one of the plurality of networking entity clusters;
select, based on a combination of the received input data, machine learning model data features based on a particular one of the KPIs to be forecasted being assigned to a particular one of the network entity clusters;
train, using the selected machine learning model data features, a machine learning model to forecast the particular KPI to be forecasted for the particular networking entity cluster; and
cause the trained machine learning model to be used for forecasting the particular KPI.
10. The apparatus as in claim 9 , wherein the networking entities comprise at least one of: a router, a switch, a wireless access point, or an access point controller.
11. The apparatus as in claim 9 , wherein the apparatus causes the trained machine learning model to be used for forecasting the particular KPI by:
deploying the trained machine learning model to one or more of the networking entities in the particular one of the networking entity clusters.
12. The apparatus as in claim 9 , wherein the apparatus causes the trained machine learning model to be used for forecasting the particular KPI by:
receiving a KPI forecast request from one of the networking entities in the particular one of the networking entity clusters;
using, in response to receiving the KPI forecast request, the trained machine learning model to forecast the particular KPI; and
providing the forecast KPI to the networking entity in the particular one of the networking entity clusters that sent the KPI forecast request.
13. The apparatus as in claim 9 , wherein the particular KPI is indicative of at least one of: a processor load, a memory load, or a traffic load.
14. The apparatus as in claim 9 , wherein the apparatus selects, based on the combination of the received input data, the machine learning model data features by:
using the entity graph to select a subset of the networking entities; and
selecting the combination of the received input data from among the subset of the networking entities.
15. The apparatus as in claim 9 , wherein the process when executed is further configured to:
use the forecast KPI to predict a tunnel failure in the network.
16. The apparatus as in claim 9 , wherein the network is a wireless network.
17. A tangible, non-transitory, computer-readable medium storing program instructions that cause a service to execute a process comprising:
receiving, at the service, input data from networking entities in a network, wherein the input data comprises synchronous time series data, asynchronous event data, and an entity graph that that indicates relationships between the networking entities in the network;
for each key performance indicator (KPI) to be forecasted among a plurality of KPIs to be forecasted, clustering, by the service, the networking entities by type in a plurality of networking entity clusters based on characteristics of each of the networking entities, such that each KPI to be forecasted is assigned to a different networking entity cluster of the plurality of networking entity clusters and each of the networking entities is capable of being included in more than one of the plurality of networking entity clusters;
selecting, by the service and based on a combination of the received input data, machine learning model data features based on a particular one of the KPIs to be forecasted being assigned to a particular one of the network entity clusters;
training, by the service and using the selected machine learning model data features, a machine learning model to forecast the particular KPI to be forecasted for the particular networking entity cluster; and
causing, by the service, the trained machine learning model to be used for forecasting the particular KPI.
18. The computer-readable medium as in claim 17 , wherein the networking entities comprise at least one of: a router, a switch, a wireless access point, or an access point controller.
19. The computer-readable medium as in claim 17 , wherein the causing of the trained machine learning model to be used for forecasting the particular KPI comprises:
deploying, by the service, the trained machine learning model to one or more of the networking entities in the particular one of the networking entity clusters.
20. The computer-readable medium as in claim 17 , wherein the causing of the trained machine learning model to be used for forecasting the particular KPI comprises:
receiving, at the service, a KPI forecast request from one of the networking entities in the particular one of the networking entity clusters;
using, by the service and in response to receiving the KPI forecast request, the trained machine learning model to forecast the particular KPI; and
providing, by the service, the forecast KPI to the networking entity in the particular one of the networking entity clusters that sent the KPI forecast request.