Predicting application and network performance

• US 10,708,152 B2
• Filed: 03/23/2017
• Issued: 07/07/2020
• Est. Priority Date: 03/23/2017
• Status: Active Grant

• Current Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Original Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Inventors: Kulshreshtha, Ashutosh, Yadav, Navindra, Deen, Khawar, Pang, Jackson, Rao, Supreeth
• Primary Examiner(s): Yohannes, Tesfay
• Application Number:
  US15/467,788
• Publication Number:
  US 20180278496A1
• Time in Patent Office: 1,202 Days
• Field of Search: 709224, 370351
• CPC Class Codes: H04L 41/083 for increasing network speed
• CPC Class Codes: H04L 41/145 involving simulating, designing, planning or modelling of a network
• CPC Class Codes: H04L 41/147 for predicting network behaviour
• CPC Class Codes: H04L 43/04 Processing captured monitoring data, e.g. for logfile generation
• CPC Class Codes: H04L 43/0852 Delays

Abstract

An application and network analytics platform can capture comprehensive telemetry from servers and network devices operating within a network. The platform can discover flows running through the network, applications generating the flows, servers hosting the applications, computing resources provisioned and consumed by the applications, and network topology, among other insights. The platform can generate various models relating one set of application and network performance metrics to another. For example, the platform can model application latency as a function of computing resources provisioned to and/or actually used by the application, its host'"'"'s total resources, and/or the distance of its host relative to other elements of the network. The platform can change the model by moving, removing, or adding elements to predict how the change affects application and network performance. In some situations, the platform can automatically act on predictions to improve application and network performance.

Claims

1. A method comprising:

capturing telemetry from a plurality of servers and a plurality of network devices of a network;

analyzing the telemetry;

generating, based on the analyzed telemetry, one or more models associating a first amount of latency of an application, one or more amounts of computing resources of a server utilized to execute the application, one or more total amounts of the computing resources, and one or more distances of the server relative to one or more of the plurality of servers and the plurality of network devices;

generating one or more updated models by applying one or more changes to the one or more models; and

determining a second amount of latency of the application using the one or more updated models;

determining the one or more changes decreases a latency of the application from the first amount of latency to the second amount of latency; and

applying the one or more changes to the network in response to the determined decrease.

2. The method of claim 1, wherein determining the one or more changes occurs in response to receiving a notification of one or more corresponding changes to the network.

3. The method of claim 1, wherein determining the one or more changes occurs at a specified interval.

4. The method of claim 1, further comprising:

determining a plurality of nodes of an application dependency map for the network by concatenating server features and process features;

determining a plurality of edges of the application dependency map based on the telemetry including one or more flows between pairs of nodes of the plurality of nodes;

determining a first feature vector for a first node of the plurality of nodes; and

determining a cluster including the first node and a second node of the plurality of nodes based on a similarity metric between the first feature vector and a second feature vector of the second node satisfying a similarity threshold.

5. The method of claim 3, further comprising:

determining a plurality of flowlets for at least one flow between the first node and the second node;

determining one or more server latencies and one or more network latencies within at least one flowlet of the plurality of flowlets;

generating a data point including the one or more server latencies and the one or more network latencies; and

updating at least one model of the one or more models using the data point.

6. The method of claim 1, further comprising:

determining at least one of an amount of CPU provisioned to or utilized to execute the application, an amount of storage provisioned to or utilized to execute the application, or an amount of disk provisioned to or utilized to execute the application; and

determining at least one of a total amount of CPU of the server, a total amount of storage of the server, or a total amount of disk of the server.

7. The method of claim 1, further comprising:

receiving the one or more distances from a customer data source or a third party data source.

8. The method of claim 1, further comprising:

determining the one or more distances using the telemetry.

9. The method of claim 1, wherein the one or more changes includes at least one of altering a configuration of a node of the network, adding a new node to the network, removing an existing node of the network, or migrating a node of the network from a first location to a second location.

10. The method of claim 1, further comprising:

determining that the application executes at a first schedule; and

determining a second schedule for the application that reduces an amount of latency of the application,

wherein the one or more changes includes changing a schedule of the application from the first schedule to the second schedule.

11. A system comprising:

a processor; and

memory including instructions that, upon execution by the processor, cause the system to;

capture telemetry from a plurality of servers and a plurality of network devices of a network;

analyze the telemetry;

generate, based on the analyzed telemetry, one or more models with one or more data points associating a first amount of latency of an application, one or more amounts of computing resources of a server utilized to execute the application, one or more total amounts of the computing resources, and one or more distances of the server relative to one or more of the plurality of servers and the plurality of network devices;

generate one or more updated models by applying one or more changes to the one or more models; and

determine a second amount of latency of the application using the one or more updated models;

determine the one or more changes decreases a latency of the application from the first amount of latency to the second amount of latency; and

apply the one or more changes to the network in response to the determined decrease.

12. The system of claim 11, wherein the instructions upon execution further cause the system to:

capture at least a portion of the telemetry at line rate from a hardware sensor embedded in an application-specific integrated circuit (ASIC) of at least one network device of the plurality of network devices.

13. The system of claim 11, wherein the instructions upon execution further cause the system to:

capture at least a portion of the telemetry from a software sensor residing within a bare metal server of the network.

14. The system of claim 11, wherein the instructions upon execution further cause the system to:

capture at least a portion of the telemetry from a plurality of software sensors residing within a plurality of virtual entities of a same physical server of the network.

15. A non-transitory computer-readable medium having instructions that, upon being executed by a processor, cause the processor to:

capture telemetry from a plurality of servers and a plurality of network devices of a network;

analyze the telemetry;

generating, based on the analyzed telemetry, one or more models associating a first amount of latency of an application, one or more amounts of computing resources of a server utilized to execute the application, one or more total amounts of the computing resources, and one or more distances of the server relative to one or more of the plurality of servers and the plurality of network devices;

generate one or more updated models by applying one or more changes to the one or more models; and

determine a second amount of latency of the application using the one or more updated models;

determine the one or more changes decreases a latency of the application from the first amount of latency to the second amount of latency; and

apply the one or more changes to the network in response to the determined decrease.

16. The non-transitory computer-readable medium of claim 15, wherein the instructions further cause the processor to:

determine the one or more changes reduces a latency of the application from the first amount of latency to the second amount of latency; and

apply one or more corresponding changes to the network in response to the determined reduction.

17. The non-transitory computer-readable medium of claim 15, wherein the instructions further cause the processor to:

generate data points for the one or more models from historical telemetry.