Workload management using a trained model
In some examples, a system creates a training data set based on features of sample workloads, the training data set comprising labels associated with the features of the sample workloads, where the labels are based on load indicators generated in a computing environment relating to load conditions of the computing environment resulting from execution of the sample workloads. The system groups selected workloads into a plurality of workload clusters based on features of the selected workloads, and computes, using a model trained based on the training data set, parameters representing contributions of respective workload clusters of the plurality of workload clusters to a load in the computing environment. The system performs workload management in the computing environment based on the computed parameters.
1. A non-transitory machine-readable storage medium comprising instructions that upon execution cause a system to:
create a training data set based on features of sample workloads, the training data set comprising labels associated with the features of the sample workloads, wherein the labels are based on load indicators generated in a computing environment relating to load conditions of the computing environment resulting from execution of the sample workloads;
group selected workloads into a plurality of workload clusters based on features of the selected workloads, wherein a workload cluster of the plurality of workload clusters comprises workloads grouped into the workload cluster according to a similarity criterion;
compute, using a model trained based on the training data set, parameters representing contributions of respective workload clusters of the plurality of workload clusters to a load condition in the computing environment, wherein the parameters comprise a first parameter representing a contribution of a first workload cluster to the load condition, and a second parameter representing a different contribution of a second workload cluster to the load condition, the first and second workload clusters being part of the plurality of workload clusters;
select a workload cluster of the plurality of workload clusters based on different values of the computed parameters; and
perform workload management in the computing environment based on the computed parameters, the workload management comprising restricting usage of a resource by a workload in the selected workload cluster.
2. The non-transitory machine-readable storage medium of claim 1 , wherein each workload of the selected workloads is performed on a respective storage volume in the computing environment.
3. The non-transitory machine-readable storage medium of claim 2 , wherein a workload cluster of the plurality of workload clusters includes workloads having similar input/output patterns.
4. The non-transitory machine-readable storage medium of claim 1 , wherein the features of the selected workloads comprise measures relating to access of resources in the computing environment.
5. The non-transitory machine-readable storage medium of claim 4 , wherein the measures relating to access of resources comprise measures relating to throughputs in writing to storage resources.
6. The non-transitory machine-readable storage medium of claim 1 , wherein the value of the parameter computed for the selected workload cluster indicates that the selected workload cluster has the workload that adversely affects a performance of a workload in another workload cluster of the plurality of workload clusters.
7. The non-transitory machine-readable storage medium of claim 1 , wherein the instructions upon execution cause the system to:
trigger the computation of the parameters representing the contributions of the respective workload clusters to the load condition in response to the load condition satisfying a criterion.
8. The non-transitory machine-readable storage medium of claim 7 , wherein the features of the selected workloads comprise measures relating to access of resources of the computing environment in corresponding time intervals, and wherein the load condition satisfies the criterion when greater than a specified quantity of the time intervals exhibit an excessive load condition.
9. The non-transitory machine-readable storage medium of claim 8 , wherein the excessive load condition is present in a given time interval of the time intervals when an acknowledgment of a completion of an access of a resource is delayed.
10. The non-transitory machine-readable storage medium of claim 1 , wherein the different values of the computed parameters represent respective different contributions of the respective workload clusters to the load condition.
11. The non-transitory machine-readable storage medium of claim 1 , wherein the instructions upon execution cause the system to:
select the workload cluster of the plurality of workload clusters based on identifying a workload cluster that has a higher contribution to the load condition than another workload cluster of the plurality of workload clusters.
12. The non-transitory machine-readable storage medium of claim 1 , wherein the resource is a shared resource shared by multiple workload clusters, and the restricting comprises a restriction on usage of the shared resource by the workload in the selected workload cluster.
13. The non-transitory machine-readable storage medium of claim 12 , wherein the different values comprise a first value of the first parameter computed for the first workload cluster, and a different second value of the second parameter computed for the second workload cluster, and wherein the first value and the second value indicate that the first workload cluster has a higher contribution to the load condition by consuming more of the shared resource than the second workload cluster.
14. The non-transitory machine-readable storage medium of claim 1 , wherein the model is a linear regression model comprising a plurality of coefficients representing the computed parameters.
15. A system comprising:
a processor; and
a non-transitory machine-readable storage medium comprising instructions executable on the processor to:
train a model using a training data set comprising information of past workloads performed in a storage environment, the training data set comprising labels associated with features of the past workloads, wherein the labels are based on load indicators generated in the storage environment relating to load conditions of the storage environment resulting from execution of the past workloads;
group current workloads into a plurality of workload clusters based on features of the current workloads, the features of the current workloads comprising measures relating to storage resource access in the storage environment, wherein a workload cluster of the plurality of workload clusters comprises workloads grouped into the workload cluster according to a similarity criterion;
compute, using the trained model, parameters representing contributions of respective workload clusters of the plurality of workload clusters to an overload condition in the storage environment, wherein the parameters comprise a first parameter representing a contribution of a first workload cluster to the overload condition, and a second parameter representing a different contribution of a second workload cluster to the overload condition, the first and second workload clusters being part of the plurality of workload clusters;
select a workload cluster of the plurality of workload clusters based on different values of the computed parameters; and
perform workload management in the storage environment based on the computed parameters, the workload management comprising restricting usage of a storage resource by a workload in the selected workload cluster.
16. The system of claim 15 , wherein the value of the parameter computed for the selected workload cluster indicates that the selected workload cluster has the workload that adversely affects a performance of a workload in another workload cluster of the plurality of workload clusters.
17. The system of claim 15 , wherein the instructions are executable on the processor to:
select the workload cluster of the plurality of workload clusters based on identifying a workload cluster that has a higher contribution to the overload condition than another workload cluster of the plurality of workload clusters.
18. The system of claim 15 , wherein the features of the selected workloads comprise measures relating to access of resources of the storage environment in corresponding time intervals, and wherein the instructions are executable on the processor to:
trigger the computation of the parameters representing the contributions of the respective workload clusters to the overload condition in response to the overload condition satisfying a criterion based on greater than a specified quantity of the time intervals exhibiting an excessive load condition, wherein the excessive load condition is present in a given time interval of the time intervals when an acknowledgment of a completion of an access of a storage resource is delayed.
19. A method of a system comprising a hardware processor, comprising:
grouping workloads into a plurality of workload clusters based on features of the workloads, wherein the features comprise measures relating to access of resources in a computing environment, and wherein a workload cluster of the plurality of workload clusters comprises workloads grouped into the workload cluster according to a similarity criterion;
training a model based on a training data set, the training data set comprising labels associated with features of sample workloads, wherein the labels are based on load indicators generated in the computing environment relating to load conditions of the computing environment resulting from execution of the sample workloads;
computing, using the model trained based on the training data set, parameters representing contributions of respective workload clusters of the plurality of workload clusters to a load in the computing environment, wherein the grouping of the workloads into the plurality of workload clusters reduces a quantity of features processed in the computing of the parameters using the model, and wherein the parameters comprise a first parameter representing a contribution of a first workload cluster to the load, and a second parameter representing a different contribution of a second workload cluster to the load, the first and second workload clusters being part of the plurality of workload clusters;
selecting a workload cluster of the plurality of workload clusters based on different values of the computed parameters; and
performing workload management in the computing environment based on the computed parameters, the workload management comprising restricting usage of a resource by a workload in the selected workload cluster.
20. The method of claim 19 , further comprising:
updating the training data set during an operation of the computing environment; and
re-training the model using the updated training data set.