Provisioning advisor
A method and system for a provisioning advisor are described which estimates the input/output operation performance of a workload on a storage system. A regression module in a provisioning advisor estimates a maximum IOPS on the storage system for buckets, or combinations of values, for various characteristics of the workloads running on the system by modeling a relationship between the workload characteristics and performance metrics gathered from the storage system. A performance module can use the estimated maximum IOPS for each bucket to update a set of working tables for the provisioning advisor, which can then be used to predict the input/output performance of a new workload to be provisioned on the storage system.
1. An adaptive workload provisioning system comprising:
a memory resource to store instructions;
one or more processors using the instructions stored in the memory resource to:
estimate a maximum number of input/output operations (IOPS) on a storage system for each of a plurality of combinations of values for a plurality of workload characteristics by modeling a relationship between the plurality of workload characteristics and one or more performance metrics observed from the storage system; wherein the plurality of workload characteristics for the combinations of values include randomness of input/output (IO) operations, IO request size, and a distribution between read and write IOs;
update a set of working tables that map each of the plurality of combinations of values to the estimated maximum IOPS for that combination of values;
predict the input/output operation performance of a new workload on the storage system using at least the set of working tables and workload characteristics for the new workload; wherein prior to commissioning the storage system, a pilot run using a subset of the plurality of workload characteristics is executed for estimating the maximum IOPS for the storage system.
2. The system of claim 1 , wherein the one or more processors use further instructions to:
create a set of master tables by estimating the maximum IOPS for a plurality of workloads on the storage system; and
copy the set of master tables into the set of working tables when the storage system is commissioned.
3. The system of claim 1 , wherein the one or more processors use further instructions to:
display the predicted input/output operation performance to a user of the storage system on a dashboard.
4. The system of claim 1 , wherein the prediction of input/output operation performance of the new workload on the storage system also considers current utilization of the storage system.
5. The system of claim 1 , wherein the storage system automatically re-provisions workloads based on predicted input/output operation performances of workloads.
6. The system of claim 1 , wherein modeling the relationship between the plurality of workload characteristics and one or more performance metrics observed from the storage system uses a robust linear regression.
7. A method of predicting input/output operation performance of a workload on a storage system, the method being implemented by one or more processors and comprising:
estimating a maximum number of input/output operations (IOPS) on the storage system for each of a plurality of combinations of values for a plurality of workload characteristics by modeling a relationship between the plurality of workload characteristics and one or more performance metrics observed from the storage system; wherein the plurality of workload characteristics for the combinations of values include randomness of input/output (IO) operations, IO request size, and a distribution between read and write IOs;
updating a set of working tables that map each of the plurality of combinations of values to the estimated maximum IOPS for that combination of values;
predicting the input/output operation performance of a new workload on the storage system using at least the set of working tables and workload characteristics for the new workload; wherein prior to commissioning the storage system, a pilot run using a subset of the plurality of workload characteristics is executed for estimating the maximum IOPS for the storage system.
8. The method of claim 7 , further comprising:
creating a set of master tables by estimating the maximum IOPS for a plurality of workloads on the storage system; and
copying the set of master tables into the set of working tables when the storage system is commissioned.
9. The method of claim 7 , further comprising:
displaying the predicted input/output operation performance to a user of the storage system on a dashboard.
10. The method of claim 7 , wherein the prediction of input/output operation performance of the new workload on the storage system also considers current utilization of the storage system.
11. The method of claim 7 , wherein the storage system automatically re-provisions workloads based on predicted input/output operation performances of workloads.
12. The method of claim 7 , wherein modeling the relationship between the plurality of workload characteristics and one or more performance metrics observed from the storage system uses a robust linear regression.
13. A non-transitory computer-readable medium that stores instructions, executable by one or more processors, to cause the one or more processors to perform operations that comprise:
estimating a maximum number of input/output operations (IOPS) on a storage system for each of a plurality of combinations of values for a plurality of workload characteristics by modeling a relationship between the plurality of workload characteristics and one or more performance metrics observed from the storage system; wherein the plurality of workload characteristics for the combinations of values include randomness of input/output (IO) operations, IO request size, and a distribution between read and write IOs;
updating a set of working tables that map each of the plurality of combinations of values to the estimated maximum IOPS for that combination of values;
predicting the input/output operation performance of a new workload on the storage system using at least the set of working tables and workload characteristics for the new workload; wherein prior to commissioning the storage system, a pilot run using a subset of the plurality of workload characteristics is executed for estimating the maximum IOPS for the storage system.
14. The non-transitory computer-readable medium of claim 13 , wherein the one or more processors use further instructions to:
create a set of master tables by estimating the maximum IOPS for a plurality of workloads on the storage system; and
copy the set of master tables into the set of working tables when the storage system is commissioned.
15. The non-transitory computer-readable medium of claim 13 , wherein the one or more processors use further instructions to:
display the predicted input/output operation performance to a user of the storage system on a dashboard.
16. The non-transitory computer-readable medium of claim 13 , wherein the prediction of input/output operation performance of the new workload on the storage system also considers current utilization of the storage system.
17. The non-transitory computer-readable medium of claim 13 , wherein the storage system automatically re-provisions workloads based on predicted input/output operation performances of workloads.
18. The non-transitory computer-readable medium of claim 13 , wherein modeling the relationship between the plurality of workload characteristics and one or more performance metrics observed from the storage system uses a robust linear regression.