Job performance breakdown
A system and method for processing application performance using application phase differentiation and detection is disclosed. Phase detection may be accomplished in a number of different ways, including by using a deterministic algorithm that looks for changes in the computing resource utilization patterns (as detected in the performance data collected). Machine learning (ML) and neural networks (e.g. sparse auto encoder SAE) may also be used. Performance data is aggregated according to phase and stored in a database along with additional application and computing system information. This database may then be used to find similar applications for performance prediction.
1. A method for processing application performance data, the method comprising:
(i) collecting, via a processor of a computational device, performance data for an application executing on a computing system;
(ii) analyzing the performance data via the processor of the computational device to identify a plurality of application phases for the application;
(iii) determining a relative weight for each of the identified application phases via the processor of the computational device;
(iv) storing aggregated performance data for the application in a database via the processor of the computational device, wherein the aggregated performance data comprises the identified application phases and their relative weights; and
(v) searching the database via the processor of the computational device to identify similar applications based on the stored aggregated performance data.
2. The method of claim 1 , wherein (ii) further comprises identifying and storing a most performance limiting process type for each of the identified application phases.
3. The method of claim 1 , wherein the identified application phases are selected from the group comprising: CPU-bound, network-bound, and I/O-bound.
4. The method of claim 1 , wherein (ii) further comprises identifying and storing a plurality of the most performance limiting process types for each of the identified application phases.
5. The method of claim 1 , further comprising repeating (i)-(vi) for a plurality of different computing systems.
6. The method of claim 5 , wherein the plurality of different computing systems comprise bare metal and virtualized cloud instances.
7. The method of claim 1 , wherein the aggregated performance data comprises a performance spectrometer for each phase.
8. The method of claim 1 , wherein (ii) is performed by applying a deterministic algorithm.
9. The method of claim 1 , wherein (ii) is performed by applying a machine learning algorithm.
10. The method of claim 1 , wherein (ii) is performed by a sparse auto encoder (SAE) neural network.
11. The method of claim 1 , further comprising configuring performance monitoring on the computer system.
12. The method of claim 1 , wherein the database includes an application performance characterization database with prior collected aggregated performance data from other applications, and wherein the similar applications include similar performance data of application phases for the other applications.
13. The method of claim 1 , wherein:
the collected performance data includes a plurality of performance counters; and
the plurality of performance counters includes at least one of:
CPU performance counters;
instructions per second counters;
cache-miss counters;
clock cycle counters; and
branch miss counters.
14. A non-transitory, computer-readable storage medium storing instructions executable by a processor of a computational device, which when executed cause the computational device to:
(i) collect performance data for an application executing on a computing system;
(ii) analyze the performance data to identify a plurality of application phases for the application;
(iii) determine a relative weight for each of the identified application phases;
(iv) store aggregated performance data for the application in a database, wherein the aggregated performance data comprises the identified application phases and their relative weights; and
(v) search the database to identify similar applications based on the stored aggregated performance data.
15. The non-transitory, computer-readable storage medium of claim 14 , wherein the instructions, when executed via the processor of the computational device, further cause the computational device to identify and store a most performance limiting process type for each of the identified application phases.
16. The non-transitory, computer-readable storage medium of claim 14 , wherein the identified application phases are selected from the group comprising: CPU-bound, network-bound, and I/O-bound.
17. The non-transitory, computer-readable storage medium of claim 14 , wherein the instructions, when executed via the processor of the computational device, further cause the computational device to repeat (i)-(vi) for a plurality of different computing systems.
18. The non-transitory, computer-readable storage medium of claim 14 , wherein the database includes an application performance characterization database with prior collected aggregated performance data from other applications, and wherein the similar applications include similar performance data of application phases for the other applications.
19. A method for processing application performance data, the method comprising:
(i) collecting, via a processor of a computational device, performance data for an application executing on a computing system;
(ii) analyzing the performance data via the processor of the computational device to identify a plurality of application phases for the application;
(iii) determining a relative weight for each of the identified application phases via the processor of the computational device;
(iv) storing aggregated performance data for the application in a database via the processor of the computational device, wherein the aggregated performance data comprises the identified application phases and their relative weights; and
(v) searching the database via the processor of the computational device to identify another application based on the stored aggregated performance data.
20. The method of claim 19 , wherein the other application includes a similar application to the application executed on the computing system.
21. The method of claim 20 , wherein the similar application is identified using the application phases of prior collected aggregated performance data from other applications that have been executed and captured in the database.
22. The method of claim 19 , wherein the plurality of application phases includes one or more of: CPU bound, memory latency bound, memory bandwidth bound, MPI/network bound, I/O latency bound, and I/O bandwidth bound.