Learning machine based computation of network join times

• US 9,553,773 B2
• Filed: 07/23/2013
• Issued: 01/24/2017
• Est. Priority Date: 02/05/2013
• Status: Active Grant

• Current Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Original Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Inventors: Vasseur, Jean-Philippe, Mermoud, Grgory, Dasgupta, Sukrit
• Primary Examiner(s): Christensen, Scott B
• Application Number:
  US13/948,367
• Publication Number:
  US 20140222975A1
• Time in Patent Office: 1,281 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: H04L 41/142 using statistical or mathematical methods
• CPC Class Codes: H04L 41/16 using machine learning or artificial intelligence

Abstract

In one embodiment, techniques are shown and described relating to learning machine based computation of network join times. In particular, in one embodiment, a device computes a join time of the device to join a computer network. During joining, the device sends a configuration request to a server, and receives instructions whether to provide the join time. The device may then provide the join time to a collector in response to instructions to provide the join time. In another embodiment, a collector receives a plurality of join times from a respective plurality of nodes having one or more associated node properties. The collector may then estimate a mapping between the join times and the node properties and determines a confidence interval of the mapping. Accordingly, the collector may then determine a rate at which nodes having particular node properties report their join times based on the confidence interval.

Claims

1. A method, comprising:

initiating joining by a device to a computer network;

during joining the computer network, sending, by the device, a configuration request to a server;

responsive to joining the computer network, computing, by the device, a join time, the join time representing an amount of time by the device to join the computer network;

receiving, at the device, instructions from the server whether to provide the join time to a collector, wherein the server is configured to determine whether or not the join time should be reported to a collector based at least in part on an optimal rate, the optimal rate only reporting the join time when the join time would provide additional information to a learning machine model executing on the collector, wherein the learning machine model calculates the optimal rate given a statistical model of a relationship between node properties and an estimate of a metric for the device; and

sending, by the device, the join time to the collector if and only if instructions from the server are received to send the join time to the collector.

2. The method as in claim 1, wherein the provided join time comprises one of either a scalar having only the provided join time or a vector having a plurality of information regarding the provided join time.

3. The method as in claim 1, wherein computing the join time comprises:

recording a wake-up time when functional elements of the device are operational; and

determining when the device is capable of sending traffic within the computer network and the difference from the recorded wake-up time.

4. The method as in claim 1, wherein the configuration request is a dynamic host configuration protocol (DHCP) request to a DHCP server.

5. The method as in claim 1, wherein the instructions comprise a bit set in a configuration reply from the server.

6. The method as in claim 1, wherein the collector is a field area router (FAR) or a network management server (NMS).

7. The method as in claim 1, wherein providing the join time comprises one of either sending an independent message or piggybacking the join time within a host message.

8. The method as in claim 7, wherein the host message is a routing protocol message.

9. The method as in claim 1, wherein the provided join time is a selection of a particular range of join times within which the computed join time falls.

10. The method as in claim 1, wherein whether to provide the join time is based on the collector:

receiving a plurality of join times from a respective plurality of nodes having one or more associated node properties;

estimating a mapping between the join times and the node properties;

determining a confidence interval of the mapping; and

determining a rate at which nodes having particular node properties report their join times based on the confidence interval.

11. A method, comprising:

receiving, by a learning machine in a computer network executing a regression computation process to estimate a mapping between join time and node properties, a plurality of join times from a respective plurality of nodes having one or more associated node properties, the join times representing an amount of time by the respective plurality of nodes to join the computer network;

estimating, by the learning machine, the mapping between the join times and the one or more associated node properties, wherein the mapping is a statistical model of a relationship between node properties and an estimate of a metric for a given node;

determining, by the learning machine, a confidence interval of the mapping;

determining, by the learning machine, a rate at which nodes having particular node properties report their join times based on the confidence interval; and

causing instructions to be sent to the respective plurality of nodes to report their join times at the determined rate.

12. The method as in claim 11, wherein the rate is each time a node joins the computer network, never, or a given rate less than each time a node joins the computer network.

13. The method as in claim 11, further comprising:

correlating node properties to particular nodes to establish an explicit list of nodes for which the rate applies.

14. The method as in claim 11, further comprising:

determining a set of nodes to explicitly probe for join times based on variation in entropy expected upon probing the set of nodes.

15. The method as in claim 11, wherein estimating the mapping and the confidence interval comprises a Variational Bayes Least Square (VBLS) regression computation.

16. An apparatus, comprising:

one or more network interfaces to communicate with a computer network;

a processor coupled to the network interfaces and adapted to execute one or more processes; and

a memory configured to store a process executable by the processor, the process when executed operable to;

initiate joining to a computer network;

during joining the computer network, send a configuration request to a server;

responsive to joining the computer network, compute a join time, the join time representing an amount of time by the apparatus to join the computer network;

receive instructions from the server whether to provide the join time to a collector, wherein the server is configured to determine whether or not the join time should be reported to a collector based at least in part on an optimal rate, the optimal rate only reporting the join time when the join time would provide additional information to a learning machine model executing on the collector, and wherein the learning machine model calculates the optimal rate given a statistical model of a relationship between node properties and an estimate of a metric for the apparatus; and

send the join time to the collector if and only if instructions from the server are received to send the join time to the collector.

17. The apparatus as in claim 16, wherein the provided join time comprises one of either a scalar having only the provided join time or a vector having a plurality of information regarding the provided join time.

18. An apparatus, comprising:

one or more network interfaces to communicate with a computer network;

a processor coupled to the network interfaces and adapted to execute one or more processes wherein at least one of the one or more processes is a learning machine process executing a regression computation process to estimate a mapping between join time and node properties; and

a memory configured to store the learning machine process executable by the processor, the learning machine process when executed operable to;

receive, as a collector in a computer network, a plurality of join times from a respective plurality of nodes having one or more associated node properties, the join times representing an amount of time by the respective plurality of nodes to join the computer network;

estimate the mapping between the join times and the one or more associated node properties, wherein the mapping is a statistical model of a relationship between node properties and an estimate of a metric for a given node;

determine a confidence interval of the mapping;

determine a rate at which nodes having particular node properties report their join times based on the confidence interval; and

cause instructions to be sent to the respective plurality of nodes to report their join times at the determined rate.

19. The apparatus as in claim 18, wherein the process when executed is further operable to:

correlate node properties to particular nodes to establish an explicit list of nodes for which the rate applies.

20. The apparatus as in claim 18, wherein the process when executed is further operable to:

determine a set of nodes to explicitly probe for join times based on variation in entropy expected upon probing the set of nodes.