Predictive analytics for network topology changes
Techniques for recommending plans to remediate a network topologies are disclosed. The techniques include logging network topology information identifying relationships between entities in a network topology over a number of time periods. The techniques also include, using the logged network topology information, predicting characteristics of the network at a future time period. The techniques further include computing a signature based on the predicted characteristics and, using the signature, determining whether the predicted characteristics meet remediation criteria. Additionally, the techniques include, in response to determining the predicted characteristics meet the remediation criteria, determining a remediation plan for the current network topology and presenting the plan to a user.
1. One or more non-transitory computer readable media comprising instructions which, when executed by one or more hardware processors, causes performance of operations comprising:
determining network topology information for a plurality of time periods, the network topology information comprising, for individual time periods of the plurality of time periods:
a plurality of entities associated with the network topology,
a plurality of relationships between the plurality of entities, and
health metrics of the plurality of entities;
storing, in a historical topology log, the network topology information for the individual time periods in association with the health metrics of the plurality of entities for the individual time periods;
based on the historical topology log, predicting characteristics of the network topology for a future time period;
computing a network topology signature based on the predicted characteristics of the network topology for the future time period;
determining whether the predicted characteristics of the network topology for the future time period meet one or more remediation criteria by applying a set of rules to the network topology signature that is based on the predicted characteristics of the network topology for the future time period;
determining a remediation plan in response to determining the predicted characteristics of the network topology for the future time period meet the one or more remediation criteria; and
presenting the remediation plan; and
generating a candidate topology that is based on the remediation plan by modifying interface elements included in a current topology based on the remediation plan.
2. The one or more non-transitory computer readable media of claim 1 , wherein predicting the characteristics of the network topology comprises:
predicting the characteristics based further on current or predicted relationships in the plurality of relationships.
3. The one or more non-transitory computer readable media of claim 1 , wherein the operations further comprise:
receiving a user input defining the future time period; and
predicting the characteristics of the network topology at the future time period by modeling behaviors of the entities associated within the network topology using network topology information for a current time period.
4. The one or more non-transitory computer readable media of claim 1 , wherein determining the remediation plan comprises:
identifying one or more candidate remediation plans; and
applying the set of rules to the one or more candidate remediation plan to select the remediation plan from the one or more candidate remediation plans.
5. The one or more non-transitory computer readable media of claim 1 , wherein the operations further comprise;
training a machine learning model to determine remediation plans for network topologies, wherein data for the training comprises: a particular network topology and a particular remediation plan for the particular network topology; and
applying the machine learning model to the network topology signature to compute the remediation plan.
6. The one or more non-transitory computer readable media of claim 1 , wherein determining the remediation plan comprises:
identifying a set of candidate remediation plans based on the network topology signature; and
determining a cost associated with the individual candidate remediation plans of set of candidate remediation plans; and
selecting the remediation plan, from the set of candidate remediation plans, based on the costs associated with the individual candidate remediation plans of the set of candidate remediation plans.
7. The one or more non-transitory computer readable media of claim 1 , wherein the health metrics comprise one or more of the following: requests per second (RPS), uptime, error rates, thread count, CPU usage, memory utilization, or disk usage, average response time, peak response time, object status, problem priority labels, number of restarts, and unbound volumes.
8. The one or more non-transitory computer readable media of claim 1 , wherein the remediation plan is selected from as set comprising:
rebooting one or more nodes, deploying a node, removing a node, allocating a node to a service, redirecting traffic to a different node, reconfiguring, a node, modifying a firewall of a node, deleting a workload, reconfiguring a workload, and scaling a workload, modifying a security policy, and generating an alert.
9. The one or more non-transitory computer readable media of claim 1 , wherein the predicted characteristics comprise information of the network topology associated with one or more of: events, anomalies, and health issues.
10. A system comprising a hardware processor and computer-readable program instructions that, when executed by the hardware processor, control the system to perform operations, comprising:
determining network topology information for a plurality of time periods, the network topology information comprising, for individual time periods of the plurality of time periods:
a plurality of entities associated with the network topology,
a plurality of relationships between the plurality of entities, and
health metrics of the plurality of entities;
storing, in a historical topology log, the network topology information for the individual time periods in association with the health metrics of the plurality of entities for the individual time periods;
based on the historical topology log, predicting characteristics of the network topology for a future time period;
computing a network topology signature based on the predicted characteristics of the network topology for the future time period;
determining whether the predicted characteristics of the network topology for the future time period meet one or more remediation criteria by applying a set of rules to a network topology signature that is based on the predicted characteristics of the network topology for the future time period;
determining a remediation plan in response to determining the predicted characteristics of the network topology for the future time period meet the one or more remediation criteria; and
presenting the remediation plan; and
generating a candidate topology that is based on the remediation plan by modifying interface elements included in a current topology based on the remediation plan.
11. The system of claim 10 , wherein predicting the characteristics of the network topology comprises:
predicting the characteristics based further on current or predicted relationships in the plurality of relationships.
12. The system of claim 10 , wherein the operations further comprise:
receiving a user input defining the future time period; and
predicting the characteristics of the network topology at the future time period by modeling behaviors of the entities associated within the network topology using network topology information for a current time period.
13. The system of claim 10 , wherein determining the remediation plan comprises:
identifying one or more candidate remediation plans; and
applying the set of rules to the one or more candidate remediation plan to select the remediation plan from the one or more candidate remediation plans.
14. The system of claim 10 , wherein the operations further comprise:
training a machine learning model to determine remediation plans for network topologies, wherein data for the training comprises: a particular network topology and a particular remediation plan for the particular network topology; and
applying the machine learning model to the network topology signature to compute the remediation plan.
15. The system of claim 10 , wherein determining the remediation plan comprises:
identifying a set of candidate remediation plans based on the network topology signature; and
determining a cost associated with the individual candidate remediation plans of set of candidate remediation plans; and
selecting the remediation plan, from the set of candidate remediation plans, based on the costs associated with the individual candidate remediation plans of the set of candidate remediation plans.
16. The system of claim 10 , wherein the health metrics comprise one or more of the following: requests per second (RPS), uptime, error rates, thread count, CPU usage, memory utilization, or disk usage, average response time, peak response time, object status, problem priority labels, number of restarts, and unbound volumes.
17. The system of claim 10 , wherein the remediation plan is selected from as set comprising:
rebooting one or more nodes, deploying a node, removing a node, allocating a node to a service, redirecting traffic to a different node, reconfiguring, a node, modifying a firewall of a node, deleting a workload, reconfiguring a workload, and scaling a workload, modifying a security policy, and generating an alert.
18. The system of claim 10 , wherein the predicted characteristics comprise information of the network topology associated with one or more of: events, anomalies, and health issues.
19. A method comprising:
determining network topology information for a plurality of time periods, the network topology information comprising, for individual time periods of the plurality of time periods:
a plurality of entities associated with the network topology,
a plurality of relationships between the plurality of entities, and
health metrics of the plurality of entities;
storing, in a historical topology log, the network topology information for the individual time periods in association with the health metrics of the plurality of entities for the individual time periods;
based on the historical topology log, predicting characteristics of the network topology for a future time period;
computing a network topology signature based on the predicted characteristics of the network topology for the future time period;
determining whether the predicted characteristics of the network topology for the future time period meet one or more remediation criteria by applying a set of rules to a network topology signature that is based on the predicted characteristics of the network topology for the future time period;
determining a remediation plan in response to determining the predicted characteristics of the network topology for the future time period meet the one or more remediation criteria; and
presenting the remediation plan; and
generating a candidate topology that is based on the remediation plan by modifying interface elements included in a current topology based on the remediation plan.
20. The method of claim 19 , wherein predicting the characteristics of the network topology comprises:
predicting the characteristics based further on current or predicted relationships in the plurality of relationships.
21. The method of claim 19 , further comprising:
receiving a user input defining the future time period; and
predicting the characteristics of the network topology at the future time period by modeling behaviors of the entities associated within the network topology using network topology information for a current time period.
22. The method of claim 19 , wherein determining the remediation plan comprises:
identifying one or more candidate remediation plans; and
applying the set of rules to the one or more candidate remediation plan to select the remediation plan from the one or more candidate remediation plans.
23. The method of claim 19 , further comprising:
training a machine learning model to determine remediation plans for network topologies, wherein data for the training comprises: a particular network topology and a particular remediation plan for the particular network topology; and
applying the machine learning model to the network topology signature to compute the remediation plan.
24. The method of claim 19 , wherein determining the remediation plan comprises:
identifying a set of candidate remediation plans based on the network topology signature; and
determining a cost associated with the individual candidate remediation plans of set of candidate remediation plans; and
selecting the remediation plan, from the set of candidate remediation plans, based on the costs associated with the individual candidate remediation plans of the set of candidate remediation plans.
25. The method of claim 19 , wherein the health metrics comprise one or more of the following: requests per second (RPS), uptime, error rates, thread count, CPU usage, memory utilization, or disk usage, average response time, peak response time, object status, problem priority labels, number of restarts, and unbound volumes.
26. The method of claim 19 , wherein the remediation plan is selected from as set comprising:
rebooting one or more nodes, deploying a node, removing a node, allocating a node to a service, redirecting traffic to a different node, reconfiguring, a node, modifying a firewall of a node, deleting a workload, reconfiguring a workload, and scaling a workload, modifying a security policy, and generating an alert.
27. The method of claim 19 , wherein the predicted characteristics comprise information of the network topology associated with one or more of: events, anomalies, and health issues.