Pattern-recognition enabled autonomous configuration optimization for data centers
A model-based approach to determining an optimal configuration for a data center may use an environmental chamber to characterize the performance of various data center configurations at different combinations of temperature and altitude. Telemetry data may be recorded from different configurations as they execute a stress workload at each temperature/altitude combination, and the telemetry data may be used to train a corresponding library of models. When a new data center is being configured, the temperature/altitude of the new data center may be used to select a pre-trained model from a similar temperature/altitude. Performance of the current configuration can be compared to the performance of the model, and if the model performs better, a new configuration based on the model may be used as an optimal configuration for the data center.
1. A non-transitory computer-readable medium comprising instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:
receiving a first performance metric from a data center, wherein:
the data center comprises a first configuration; and
the first performance metric is generated as the data center executes a workload;
receiving one or more environmental characteristics that are measured in an environment surrounding the data center as the data center executes the workload;
identifying a model that was trained using data from one or more data center configurations operating in an environment having environmental characteristics that are similar to the one or more environmental characteristics that are measured in the environment surrounding the data center as the data center executes the workload;
generating a second performance metric from the model;
comparing the second performance metric to the first performance metric; and
determining a second configuration for the data center from the model.
2. The non-transitory computer-readable medium according to claim 1 , wherein the model is identified from among a plurality of models, wherein each of the plurality of models is trained using data from one or more data center configurations operating in environments having different environmental characteristics.
3. The non-transitory computer-readable medium according to claim 1 , wherein the one or more data center configurations are placed in an environmental chamber to control the environmental characteristics while the one or more data center configurations execute the workload.
4. The non-transitory computer-readable medium according to claim 3 , wherein the environmental chamber controls a temperature and a simulated altitude around the one or more data center configurations.
5. The non-transitory computer-readable medium according to claim 4 , wherein the environmental chamber is configured to simulate an altitude and collect a full telemetry suite of signals at a plurality of temperatures at the altitude.
6. The non-transitory computer-readable medium according to claim 5 , wherein the plurality of temperatures includes temperatures between approximately 15° C. and 35° C.
7. The non-transitory computer-readable medium according to claim 4 , wherein the environmental chamber simulates a plurality of altitudes between approximately sea level and 5000 feet.
8. The non-transitory computer-readable medium according to claim 7 , wherein the plurality of temperatures are incremented at intervals of approximately 1° C.
9. The non-transitory computer-readable medium according to claim 1 , wherein the one or more environmental characteristics comprise an ambient temperature surrounding the data center and an altitude at which the data center is installed.
10. The non-transitory computer-readable medium according to claim 1 , wherein identifying the model that was trained using data from one or more data center configurations operating in the environment having environmental characteristics that are similar to the one or more environmental characteristics comprises:
executing a nearest-neighbor algorithm to identify the environmental characteristics that are most similar to the one or more environmental characteristics.
11. The non-transitory computer-readable medium according to claim 1 , wherein the nearest-neighbor algorithm minimizes a difference between a temperature for the model and a temperature for the data center, and minimizes a difference between a simulated altitude of the model and an altitude of the data center.
12. The non-transitory computer-readable medium according to claim 1 , wherein the first configuration comprises a number and type of processors in the data center.
13. The non-transitory computer-readable medium according to claim 1 , wherein the first configuration comprises a number and type of hard disk drives in the data center.
14. The non-transitory computer-readable medium according to claim 1 , wherein the first configuration comprises a data cache size.
15. The non-transitory computer-readable medium according to claim 1 , wherein the data center comprises a cloud data center.
16. The non-transitory computer-readable medium according to claim 1 , wherein the operations further comprise:
comparing the first performance metric to a Service Level Agreement (SLA); and
determining that the first performance metric does not meet the SLA, wherein the model is identified in response to determining that the first performance metric does not meet the SLA.
17. The non-transitory computer-readable medium according to claim 1 , wherein the operations further comprise:
determining that the second performance metric exceeds the first performance metric; and
providing the second configuration to be implanted by the data center.
18. The non-transitory computer-readable medium according to claim 1 , wherein the first performance metric comprises a processor performance for each processor in the data center, and an I/O performance for each hard disk drive in the data center.
19. A system comprising:
one or more processors; and
one or more memory devices comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
receiving a first performance metric from a data center, wherein:
the data center comprises a first configuration; and
the first performance metric is generated as the data center executes a workload;
receiving one or more environmental characteristics that are measured in an environment surrounding the data center as the data center executes the workload;
identifying a model that was trained using data from one or more data center configurations operating in an environment having environmental characteristics that are similar to the one or more environmental characteristics that are measured in the environment surrounding the data center as the data center executes the workload;
generating a second performance metric from the model;
comparing the second performance metric to the first performance metric; and
determining a second configuration for the data center from the model.
20. A method for autonomously determining an optimal configuration for data centers using a library of models pre-trained at various combinations of environmental characteristics, the method comprising:
receiving a first performance metric from a data center, wherein:
the data center comprises a first configuration; and
the first performance metric is generated as the data center executes a workload;
receiving one or more environmental characteristics that are measured in an environment surrounding the data center as the data center executes the workload;
identifying a model that was trained using data from one or more data center configurations operating in an environment having environmental characteristics that are similar to the one or more environmental characteristics that are measured in the environment surrounding the data center as the data center executes the workload;
generating a second performance metric from the model;
comparing the second performance metric to the first performance metric; and
determining a second configuration for the data center from the model.