Obtaining of statistical performance of storage system
Obtaining or facilitating obtaining statistical performance of a storage system is described. For instance, a method comprises: dividing a value range of performance of a storage system into a plurality of adjacent sub-ranges, an upper limit value of each sub-range being a predetermined multiple of a non-zero lower limit value of the sub-range. The method also comprises: determining a distribution of a plurality of measurement values of the performance among the plurality of sub-ranges. The method further comprises: estimating, based on the distribution, a percentile of the plurality of measurement values corresponding to a predetermined percentage to indicate statistical performance of the storage system. As a result, statistical performance indicators of the storage system can be obtained with an acceptable error using a small amount of memory resources.
1. A computer-implemented method, comprising:
obtaining, by a system comprising a processor, a plurality of measurement values of performance of a storage system;
dividing, by the system, a value range of the performance of the storage system into a plurality of adjacent sub-ranges, a respective upper limit value of each sub-range of the plurality of adjacent sub-ranges being a predetermined multiple of a respective non-zero lower limit value of the sub-range;
determining, by the system, a distribution of the plurality of measurement values of the performance among the plurality of adjacent sub-ranges;
estimating, by the system and based on the distribution, a percentile of the plurality of measurement values corresponding to a predetermined percentage to indicate a statistical performance of the storage system; and
determining, by the system, a performance issue of the storage system based on the percentile of the plurality of measurement values.
2. The method of claim 1 , wherein the dividing the value range into the plurality of adjacent sub-ranges comprises:
determining a tolerable relative error for estimating the percentile;
determining the predetermined multiple based on the tolerable relative error; and
determining, based on the value range and the predetermined multiple, the respective upper limit value and the respective non-zero lower limit value for each sub-range.
3. The method of claim 1 , wherein the dividing the value range into the plurality of adjacent sub-ranges comprises:
rounding the respective upper limit value and the respective non-zero lower limit value for each sub-range; and
combining a group of the plurality of adjacent sub-ranges having a same upper limit value and a same lower limit value after rounding into a single sub-range.
4. The method of claim 1 , wherein the dividing the value range into the plurality of adjacent sub-ranges comprises:
in response to a lower limit value of a sub-range of the plurality of adjacent sub-ranges being determined to be zero, setting an upper limit value of the sub-range to a predetermined non-zero value.
5. The method of claim 1 , wherein the determining the distribution comprises:
for each measurement value of the plurality of measurement values,
determining a sub-range including the measurement value from the plurality of adjacent sub-ranges; and
increasing incrementally a counter associated with the sub-range, the counter being used for recording a number of measurement values of the plurality of measurement values included in the sub-range.
6. The method of claim 5 , wherein the determining the sub-range comprises:
searching the sub-range using a binary search algorithm.
7. The method of claim 1 , wherein the estimating the percentile comprises:
determining a number of the plurality of measurement values;
determining, based on the number and the predetermined percentage, a sequence number of the percentile in the plurality of measurement values;
determining, from the plurality of adjacent sub-ranges, a target sub-range including the percentile based on the sequence number; and
obtaining an estimated value of the percentile based on an upper limit value and a lower limit value of the target sub-range.
8. The method of claim 7 , wherein the determining the number of the plurality of measurement values comprises:
determining a plurality of count values of a plurality of counters respectively associated with the plurality of adjacent sub-ranges, the plurality of counters being respectively used for recording the numbers of measurement values included in the plurality of adjacent sub-ranges; and
obtaining the number of the plurality of measurement values based on the plurality of count values.
9. The method of claim 7 , wherein the obtaining the estimated value comprises:
calculating an average value of the upper limit value and the lower limit value of the target sub-range as the estimated value.
10. An electronic device, comprising:
at least one processor; and
at least one memory storing computer program instructions, the at least one memory and the computer program instructions being configured, with the at least one processor, to cause the electronic device to:
obtain a plurality of measurement values of performance of a storage system;
divide a value range of the performance of the storage system into a plurality of adjacent sub-ranges, a respective upper limit value of each sub-range being a predetermined multiple of a respective non-zero lower limit value of the sub-range;
determine a distribution of the plurality of measurement values of the performance among the plurality of adjacent sub-ranges;
estimate, based on the distribution, a percentile of the plurality of measurement values corresponding to a predetermined percentage to indicate a statistical performance of the storage system; and
determine a performance issue of the storage system based on the percentile of the plurality of measurement values.
11. The electronic device of claim 10 , wherein the at least one memory and the computer program instructions are configured, with the at least one processor, to cause the electronic device to divide the value range into the plurality of adjacent sub-ranges by:
determining a tolerable relative error for estimating the percentile;
determining the predetermined multiple based on the tolerable relative error; and
determining, based on the value range and the predetermined multiple, the respective upper limit value and the respective non-zero lower limit value for each sub-range.
12. The electronic device of claim 10 , wherein the at least one memory and the computer program instructions are configured, with the at least one processor, to cause the electronic device to divide the value range into the plurality of adjacent sub-ranges by:
rounding the respective upper limit value and the respective non-zero lower limit value for each sub-range; and
combining a plurality of adjacent sub-ranges having a same respective upper limit value and a same respective non-zero lower limit value after rounding into a single sub-range.
13. The electronic device of claim 10 , wherein the at least one memory and the computer program instructions are configured, with the at least one processor, to cause the electronic device to divide the value range into the plurality of adjacent sub-ranges by:
in response to a corresponding lower limit value of a selected sub-range of the plurality of adjacent sub-ranges being determined to be zero, setting the respective upper limit value of the selected sub-range to a predetermined non-zero value.
14. The electronic device of claim 10 , wherein the at least one memory and the computer program instructions are configured, with the at least one processor, to cause the electronic device to determine the distribution by:
for each of the plurality of measurement values,
determining a sub-range including the measurement value from the plurality of adjacent sub-ranges, the determining resulting in a determined sub-range; and
increasing incrementally a counter associated with the determined sub-range, the counter being used for recording a number of measurement values included in the determined sub-range.
15. The electronic device of claim 14 , wherein the at least one memory and the computer program instructions are configured, with the at least one processor, to cause the electronic device to determine the sub-range by:
searching the sub-range using a binary search algorithm, resulting in the determined sub-range.
16. The electronic device of claim 10 , wherein the at least one memory and the computer program instructions are configured, with the at least one processor, to cause the electronic device to estimate the percentile by:
determining a number of the plurality of measurement values;
determining, based on the number and the predetermined percentage, a sequence number of the percentile in the plurality of measurement values;
determining, from the plurality of adjacent sub-ranges, a target sub-range including the percentile based on the sequence number; and
obtaining an estimated value of the percentile based on a corresponding upper limit value and a corresponding lower limit value of the target sub-range.
17. The electronic device of claim 16 , wherein the at least one memory and the computer program instructions are configured, with the at least one processor, to cause the electronic device to determine the number of the plurality of measurement values by:
determining a plurality of count values of a plurality of counters respectively associated with the plurality of adjacent sub-ranges, the plurality of counters being respectively used for recording numbers of measurement values included in the plurality of adjacent sub-ranges; and
obtaining the number of the plurality of measurement values based on the plurality of count values.
18. The electronic device of claim 16 , wherein the at least one memory and the computer program instructions are configured, with the at least one processor, to cause the electronic device to obtain the estimated value by:
calculating an average value of the corresponding upper limit value and the corresponding lower limit value of the target sub-range as the estimated value.
19. A computer program product being tangibly stored on a non-volatile computer-readable medium and comprising machine-executable instructions which, when executed, causing a machine to perform operations, comprising:
obtaining, by a system comprising a processor, a plurality of measurement values of performance of a storage system;
dividing a value range of the performance of the storage system into adjacent sub-ranges, an upper limit value of each sub-range being a defined multiple of a non-zero lower limit value of the sub-range;
determining a distribution of the plurality of measurement values of the performance among the adjacent sub-ranges;
estimating, based on the distribution, a percentile of the measurement values corresponding to a predetermined percentage to indicate a statistical performance of the storage system; and
determining, by the system, a performance issue of the storage system based on the percentile of the plurality of measurement values.
20. The computer program product of claim 19 , wherein the dividing the value range into the adjacent sub-ranges comprises:
determining a tolerable relative error for estimating the percentile;
determining the defined multiple based on the tolerable relative error; and determining, based on the value range and the defined multiple, the upper limit value and the non-zero lower limit value for each sub-range.