Resource cache classification using machine learning

1. A method, comprising:

determining, by a processor, a plurality of resources loaded during rendering of a web page on a client device;

determining, by the processor, a duration of time taken for each of the plurality of resources to fully load;

clustering, by the processor, the plurality of resources into clusters, comprising:

a first cluster consisting of resources having the shortest durations of the plurality of resources; and

a second cluster consisting of resources having the longest durations of the plurality of resources;

classifying, by the processor, those resources of the first cluster as cached resources; and

classifying, by the processor, those resources of the second cluster as non-cached resources.

2. The method as in claim 1 , wherein clustering the plurality of resources into clusters further comprises a third cluster consisting of resources of the plurality of resources having durations between the shortest durations and the longest durations, the method further comprising:

classifying those resources of the third cluster as unclassified resources.

3. The method as in claim 1 , further comprising:

determining that the rendering of the web page is the first time the web page is rendered on the client device; and

classifying all of the plurality of resources as non-cached resources in response to the rendering of the web page being the first time the web page is rendered on the client device.

4. The method as in claim 3 , wherein determining that the rendering of the web page is the first time the web page is rendered on the client device is in response to:

determining that a given percentage of the plurality of resources have a duration beyond a given threshold.

5. The method as in claim 4 , wherein the given threshold is based on which web browser performs the rendering of the web page.

6. The method as in claim 1 , further comprising:

determining a proportion of cached resources and non-cached resources; and

displaying the proportion within a web page performance dashboard.

7. The method as in claim 1 , further comprising:

correlating a plurality of snapshots of renderings of the web page from a plurality of client devices; and

generating a report on cached and non-cached resources across the plurality of snapshots of renderings of the web page.

8. The method as in claim 1 , wherein the plurality of resources comprise both same-origin and cross-origin resources.

9. The method as in claim 1 , wherein clustering the plurality of resources into clusters is based on a machine learning algorithm.

10. A tangible, non-transitory, computer-readable medium storing program instructions that cause a computer to execute a process comprising:

determining a plurality of resources loaded during rendering of a web page on a client device;

determining a duration of time taken for each of the plurality of resources to fully load;

clustering the plurality of resources into clusters, comprising:

a first cluster consisting of resources having the shortest durations of the plurality of resources; and

a second cluster consisting of resources having the longest durations of the plurality of resources;

classifying those resources of the first cluster as cached resources; and

classifying those resources of the second cluster as non-cached resources.

11. The computer-readable medium as in claim 10 , wherein clustering the plurality of resources into clusters further comprises a third cluster consisting of resources of the plurality of resources having durations between the shortest durations and the longest durations, the process further comprising:

classifying those resources of the third cluster as unclassified resources.

12. The computer-readable medium as in claim 10 , the process further comprising:

determining that the rendering of the web page is the first time the web page is rendered on the client device; and

classifying all of the plurality of resources as non-cached resources in response to the rendering of the web page being the first time the web page is rendered on the client device.

13. The computer-readable medium as in claim 12 , wherein determining that the rendering of the web page is the first time the web page is rendered on the client device is in response to:

determining that a given percentage of the plurality of resources have a duration beyond a given threshold.

14. The computer-readable medium as in claim 13 , wherein the given threshold is based on which web browser performs the rendering of the web page.

15. The computer-readable medium as in claim 10 , the process further comprising:

determining a proportion of cached resources and non-cached resources; and

displaying the proportion within a web page performance dashboard.

16. The computer-readable medium as in claim 10 , the process further comprising:

correlating a plurality of snapshots of renderings of the web page from a plurality of client devices; and

generating a report on cached and non-cached resources across the plurality of snapshots of renderings of the web page.

17. The computer-readable medium as in claim 10 , wherein the plurality of resources comprise both same-origin and cross-origin resources.

18. The computer-readable medium as in claim 10 , wherein clustering the plurality of resources into clusters is based on a machine learning algorithm.

19. An apparatus, comprising:

one or more network interfaces configured to communicate in 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:

determine a plurality of resources loaded during rendering of a web page on a client device;

determine a duration of time taken for each of the plurality of resources to fully load;

cluster the plurality of resources into clusters, comprising:

a first cluster consisting of resources having the shortest durations of the plurality of resources; and

a second cluster consisting of resources having the longest durations of the plurality of resources;

classify those resources of the first cluster as cached resources; and

classify those resources of the second cluster as non-cached resources.

20. The apparatus as in claim 19 , wherein clustering the plurality of resources into clusters further comprises a third cluster consisting of resources of the plurality of resources having durations between the shortest durations and the longest durations, wherein the process when executed is further operable to:

classify those resources of the third cluster as unclassified resources.

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

determine that the rendering of the web page is the first time the web page is rendered on the client device; and

classify all of the plurality of resources as non-cached resources in response to the rendering of the web page being the first time the web page is rendered on the client device.