Predictive rollup and caching for application performance data

• US 10,404,822 B2
• Filed: 04/29/2016
• Issued: 09/03/2019
• Est. Priority Date: 04/29/2016
• Status: Active Grant

• Current Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Original Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Inventors: Whitney, Jonathan, Lu, Zhijiang, Rusin, Rafal
• Primary Examiner(s): Chang, Li Wu
• Application Number:
  US15/142,882
• Publication Number:
  US 20170316321A1
• Time in Patent Office: 1,222 Days
• Field of Search: None
• CPC Class Codes: G06F 16/9574 of access to content, e.g. by caching
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: H04L 67/5681 Pre-fetching or pre-delivering data based on network characteristics

Abstract

In one aspect, a system for pre-fetching performance data in a monitored environment is disclosed. The system can include a processor; a memory; and one or more modules stored in the memory. The one or more modules are executable by the processor to perform operations including: record queries that request for application performance data with latencies longer than a threshold; learn access patterns in the recorded queries with latencies longer than the threshold; pre-fetch and cache the application performance data requested by the recorded queries before the same recorded queries are requested next time; and provide the pre-fetched application performance data from the cache when the same recorded queries are requested next time.

Claims

1. A system for pre-fetching performance data in a monitored environment, including:

a processor;

a memory; and

one or more modules stored in the memory and executable by the processor to perform operations including;

record queries that request application performance data with latencies longer than a threshold;

learn, via a machine learning process, access patterns in the recorded queries with latencies longer than the threshold, wherein the access pattern includes when and how often a same query is requested;

based on the learned access patterns, pre-fetch and cache the application performance data requested by the recorded queries before the same recorded queries are requested next time, wherein prefetching includes;

identifying a roll up process required by the recorded queries, wherein the roll up process includes processing data into a next granular format; and

provide the pre-fetched application performance data from the cache when the same recorded queries are requested next time.

2. The system of claim 1, wherein the one or more modules stored in the memory and executable by the processor to perform operations including:

determine newly streamed performance data is relevant to the pre-fetched and cached performance data; and

append the pre-fetched and cached data with the newly streamed performance data that are determined to be relevant.

3. The system of claim 1, wherein the one or more modules stored in the memory and executable by the processor to pre-fetch and cache the application performance data by performing operations including:

identify a type of performance data; and

pre-processing the identified type of performance data using the identified roll up process.

4. The system of claim 3, wherein the roll up process includes raw data received over a predetermined time interval.

5. The system of claim 1, wherein the access pattern includes a predetermined time period access.

6. The system of claim 1, wherein the application performance data includes time needed to complete a business transaction.

7. The system of claim 1, wherein the application performance data includes time need to access a database.

8. The system of claim 1, wherein the application performance data includes a resource usage rate.

9. A method for pre-fetching performance data in a monitored environment, including:

receiving a query for application performance data associated with the monitored environment;

determining whether latency of the received query exceeds a threshold;

recording the query based on the determining that the latency of the received query exceeds the threshold;

learning, via a machine learning process, an access pattern in the recorded query with the latency longer than the threshold, wherein the access pattern includes when and how often a same query is requested;

based on the learned access patterns, pre-fetching and caching the application performance data requested by the recorded query before the same recorded query is requested next time wherein prefetching includes;

identifying a roll up process required by the recorded queries, wherein the roll up process includes processing data into a next granular format;

providing the pre-fetched application performance data from the cache when the same recorded query is requested next time; and

updating the learned access pattern based on data obtained during the next time when the same recorded query is requested.

10. The method of claim 9, including:

determining whether newly streamed performance data are relevant to the pre-fetched and cached performance data; and

appending the pre-fetched and cached data with the newly streamed performance data that are determined to be relevant.

11. The method of claim 9, including:

identifying a type of performance data; and

pre-processing the identified type of performance data using the identified roll up process.

12. The method of claim 11, wherein the roll up process includes raw data in minute interval, average hourly data, average daily data, average weekly data, average monthly data, or average yearly data.

13. The method of claim 9, wherein the access pattern includes a predetermined time period access.

14. The method of claim 9, wherein the application performance data includes time needed to complete a business transaction.

15. The method of claim 9, wherein the application performance data includes time need to access a database.

16. The method of claim 9, wherein the application performance data includes a resource usage rate.

17. A non-transitory computer readable medium embodying instructions when executed by a processor to cause operations to be performed including:

recording a query requesting performance data that has a latency that exceeds a threshold latency time;

learning, by a machine learning process, an access pattern in the recorded query with the latency longer than the threshold, wherein the access pattern includes when and how often a same query is requested;

based on the learned access patterns, pre-fetching and caching the application performance data requested by the recorded query before the same recorded query is requested next time wherein prefetching includes;

identifying a roll up process required by the recorded queries, wherein the roll up process includes processing data into a next granular format; and

providing the pre-fetched application performance data from the cache when the same recorded query is requested next time.

18. The non-transitory computer readable medium of claim 17, wherein the instructions when executed by the processor can cause operations to be performed including:

updating the learned access pattern based on data obtained during the next time when the same recorded query is requested.

19. The non-transitory computer readable medium of claim 17, wherein the instructions when executed by the processor can cause operations to be performed including:

determining whether newly streamed performance data are relevant to the pre-fetched and cached performance data; and

appending the pre-fetched and cached data with the newly streamed performance data that are determined to be relevant.

20. The non-transitory computer readable medium of claim 17, wherein the instructions when executed by the processor can cause operations to be performed including:

identifying a type of performance data; and

pre-processing the identified type of performance data using the identified roll up process.

21. The non-transitory computer readable medium of claim 17, wherein the roll up process includes raw data in a time interval resolution or averaged data over a predetermined time interval.

22. The non-transitory computer readable medium of claim 17, wherein the access pattern includes a predetermined time interval access.