Data assurance in server consolidation

• US 20080027961A1
• Filed: 07/28/2006
• Published: 01/31/2008
• Est. Priority Date: 07/28/2006
• Status: Active Grant

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Rolia, Jerome, Farkas, Keith I., Fremont, Marie-Jo, Arlitt, Martin F.
• Granted Patent: US 8,051,162 B2
• Time in Patent Office: Days
• US Class Current: 707/101
• CPC Class Codes: G06F 11/3409 for performance assessment
• CPC Class Codes: G06F 11/3452 Performance evaluation by statistical analysis
• CPC Class Codes: G06F 11/3476 Data logging G06F11/14, G06F11/2205 take precedence
• CPC Class Codes: G06F 11/3495 for systems
• CPC Class Codes: G06F 2201/81 Threshold
• CPC Class Codes: G06F 2201/835 Timestamp
• CPC Class Codes: G06F 2201/88 Monitoring involving counting
• CPC Class Codes: H04L 41/0681 Configuration of triggering conditions
• CPC Class Codes: H04L 41/5009 Determining service level performance parameters or violations of service level contracts, e.g. violations of agreed response time or mean time between failures [MTBF]
• CPC Class Codes: H04L 43/022 by sampling
• CPC Class Codes: H04L 43/08 Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
• CPC Class Codes: H04L 43/106 using time related information in packets, e.g. by adding timestamps
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• CPC Class Codes: H04L 67/1008 based on parameters of servers, e.g. available memory or workload monitoring of computer activity G06F11/30
• CPC Class Codes: H04L 67/1012 based on compliance of requirements or conditions with available server resources
• CPC Class Codes: H04L 67/1029 using data related to the state of servers by a load balancer

Abstract

A method for data assurance in server consolidation is provided. The method includes collecting an inventory of a plurality of source servers in a desired environment and performance data of such source servers, evaluating and checking a data structure of the performance data, applying predetermined time stamps, checks, and statistic computations to the performance data, and evaluating a data quality of the performance data.

Claims

1. A method for data assurance in server consolidation, comprising:

collecting performance data of a plurality of source servers in a desired environment that are selected for server consolidation, the performance data includes an input trace of each of the plurality of source servers, the input trace includes data of a plurality of performance metrics collected over a predetermined period of time at predetermined time intervals for the each performance metric;

evaluating and checking a data structure of each of the input traces;

applying with a computerized system predetermined time stamps, predetermined checks, and predetermined statistic computations to each of the input traces; and

evaluating a data quality of the input traces based on the applying with the computerized system.

2. The computer-implemented method of claim 1, wherein evaluating and checking the data structure comprises:

checking meta information of each of the input traces to ensure correct types of data;

checking each of the input traces to exclude any one of the plurality of servers that has an associated one of the input traces with at least one missing performance metrics from the plurality of performance metrics; and

checking a time series alignment of each of the input traces by comparing a begin time and an end time of each of the input traces with a predetermined begin time and a predetermined end time of an expected output trace.

3. The method of claim 1, wherein applying with a computerized system comprises:

examining the plurality of performance metrics in each of the predetermined time intervals in each of the input traces for predetermined errors;

for each of the input traces, removing any of the plurality of predetermined time intervals, and the plurality of performance metrics therein, from the each input trace when the any predetermined time interval is out of order with respect to an adjacent previous time interval in the plurality of predetermined time intervals;

checking a time stamp misalignment of each of the input traces by comparing time stamps for the predetermined time intervals in the each input trace with predetermined time stamps in an expected output trace;

determining whether the each input trace includes data for the plurality of performance metrics in an extra time interval having a time stamp not accounted by the predetermined time stamps in the expected output trace; and

performing with the computerized system removal of invalid data points and computation of statistics of each of the performance metrics in the each input trace, wherein each of the data points in the input trace represents a data value of one of the plurality of performance metrics at one of the predetermined time intervals in the each input trace.

4. The method of claim 3, wherein performing with the computerized system removal of invalid data points and computation of statistics comprises:

determining a number of NULL values for each of the plurality of performance metrics in each of the input traces;

determining a number of data points for each of the plurality of performance metrics in the each input trace that are valid because the valid data points have values within a predetermined range;

determining a number of data values for each of the plurality of performance metrics in the each input trace that are invalid because the invalid data points have values outside the predetermined range;

for each of the invalid data points determined, generating an error message to report the each invalid data point; and

for each of the invalid data points determined, assigning a predetermined value to the each data point.

5. The method of claim 1, further comprising:

generating any missing data from the input traces that result from the applying with the computerized system; and

evaluating the data quality of the input traces further based on the any generated data;

wherein the step of generating any missing data from the input traces includes,

a) calculating a time difference between two adjacent ones of the predetermined time intervals for each of the input traces;

p2 b) determining whether the time difference is greater than a predetermined time difference between time intervals in an expected output trace;

c) upon a determination that the time difference is greater than the predetermined time difference, calculating a number of the predetermined time intervals with missing data for the each input trace;

d) generating an error message indicating the number of the predetermined time intervals with missing data for the each input trace, a begin time for the number, and an end time for the number;

e) generating new data for the missing data for the each performance metric based on predetermined criteria;

f) calculating a missing-row number as a number of groups of consecutive ones of the predetermined time intervals with missing data for the each performance metric;

g) repeating a) to f) through a last one of the predetermined time intervals in the each input trace; and

h) checking whether there is missing data for the each performance metric at a beginning or an end of the each input trace.

6. The method of claim 4, wherein evaluating a data quality of the input traces comprises:

calculating a ratio of the number of NULL values to the number of valid data points for each of the plurality of performance metrics in the each input trace;

determining whether the calculated ratio is greater than a predetermined ratio value and the each performance metric is required; and

upon a determination that the calculated ratio is greater than the predetermined ratio value and the each performance metric is required, generating an error message and marking the server from which the each input trace is collected with an Unchanged status.

7. The method of claim 4, wherein evaluating a data quality of the input traces comprises:

calculating a ratio of the number of NULL values to the number of valid data points for each of the plurality of performance metrics in the each input trace;

determining whether the calculated ratio is greater than a predetermined ratio value and the each performance metric is optional; and

upon a determination that the calculated ratio is greater than the predetermined ratio value and the each performance metric is optional, generating an error message and recommending the server from which the each input trace is collected be marked with an Unchanged status.

8. The method of claim 5, wherein evaluating a data quality of the input traces comprises:

determining whether the missing-row number is greater than a predetermined missing-row threshold value;

upon a determination that the missing-row number is greater than the predetermined missing-row threshold value, generating an error message and recommending the server from which the each input trace is collected be marked with an Unchanged status.

9. The method of claim 3, evaluating a data quality of the input traces comprises:

determining whether a number of extra rows obtained from the step of determining whether the each input trace includes data in the extra time interval is greater than a predetermined extra-rows-error threshold value; and

upon a determination that the number of extra rows is greater than the predetermined extra-rows-error threshold value, generating an error message and marking the server from which the each input trace is collected with an Unchanged status.

10. The method of claim 4, wherein the step of evaluating a data quality of the input traces comprises:

calculating a ratio of the number of invalid data points to the number of valid data points for each of the plurality of performance metrics in the each input trace;

determining whether the calculated ratio is greater than predetermined ratio value;

upon a determination that the calculated ratio is greater than the predetermined ratio value, generating an error message and marking a server from which the each input trace is collected with an Unchanged status.

11. The method of claim 4, wherein evaluating a data quality of the input traces comprises:

calculating a percentage of data points for each of the plurality of performance metrics in the each input trace that exceeds a predetermined value;

determining whether the calculated percentage is greater than a predetermined percentage value;

upon a determination that the calculated percentage is greater than the predetermined percentage value, generating an error message and requesting user input on whether a server from which the each input trace is collected is to be marked with an Unchanged status.

12. A computer-readable medium on which is encoded program code, the program code comprising:

program code for accessing performance data of a plurality of source servers, the performance data includes an input trace of each of the plurality of source servers, the input trace includes data of a plurality of predetermined performance metrics arranged as a plurality of data sets, each of the plurality of data sets includes accessed values of the plurality of performance metrics at a particular instance;

program code for evaluating a data structure of each of the input traces by checking metadata of each of the input traces to ensure that the predetermined performance metrics are found in each of the input traces;

program code for applying to each of the input traces one of;

predetermined time stamps, predetermined checks, and predetermined statistical computations;

program code for generating any missing data from each of the input traces that result from running the program code for applying to each of the input traces; and

program code for evaluating a data quality of each of the input traces with at least any generated data from running the program code for generating.

13. The computer readable medium of claim 12, wherein the program code for evaluating further comprises:

program code for excluding any one of the plurality of servers that has an associated one of the input traces with a missing one of the predetermined performance metrics; and

program code for checking a time series alignment of each of the input traces by comparing a begin time and an end time of each of the input traces with a predetermined begin time and a predetermined end time of an expected output trace.

14. The computer readable medium of claim 12, further comprising:

program code for concurrently processing the program code for evaluating the data structure of each of the input traces, the program code for applying to each of the input traces, the program code for generating any missing data from each of the input traces, and program code for evaluating the data quality of each of the input traces.

15. The computer readable medium of claim 12, wherein the program code for applying to each of the input traces comprises:

program code for examining the plurality of performance metrics in each of the predetermined time intervals in each of the input traces for predetermined errors;

program code for, for each of the input traces, removing any of the plurality of predetermined time intervals, and the plurality of performance IS metrics therein, from the each input trace when the any predetermined time interval is out of order with respect to an adjacent previous time interval in the plurality of predetermined time intervals;

program code for checking a time stamp misalignment of each of the input traces by comparing time stamps for the predetermined time intervals in the each input trace with predetermined time stamps in an expected output trace;

program code for determining whether the each input trace includes data for the plurality of performance metrics in an extra time interval having a time stamp not accounted by the predetermined time stamps in the expected output trace; and

program code for performing removal of invalid data points, based on expected minimum and maximum values as found in the metadata, wherein each of the data points in the input trace represents a data value of one of the plurality of performance metrics at one of the predetermined time intervals in the each input trace.

16. The computer readable medium of claim 12, wherein the program code for generating any missing data from the input traces comprises:

a) program code for calculating a time difference between two adjacent ones of the predetermined time intervals for each of the input traces;

b) program code for determining whether the time difference is greater than a predetermined time difference between time intervals in an expected output trace;

c) program code for, upon a determination that the time difference is greater than the predetermined time difference, calculating a number of the predetermined time intervals with missing data for the each input trace;

d) program code for generating an error message indicating the number of the predetermined time intervals with missing data for the each input trace, a begin time for the number, and an end time for the number;

e) program code for generating new data for the missing data for the each performance metric based on predetermined criteria;

f) program code for calculating a missing-row number as a number of groups of consecutive ones of the predetermined time intervals with missing data for the each performance metric;

g) program code for repeating program codes in a) through f) through a last one of the predetermined time intervals in the each input trace; and

h) program code for checking whether there is missing data for the each performance metric at a beginning or an end of the each input trace.

17. A method for providing a server consolidation service to a customer, comprising:

accessing performance data of a plurality of source servers in a desired environment, the performance data originating from the customer and including an input trace of each of the plurality of source servers, the input trace includes data of a plurality of predetermined performance metrics arranged as a plurality of data sets, each of the plurality of data sets includes accessed values of the plurality of performance metrics at a particular instance;

assessing a quality of the performance data and cleaning the performance data;

providing the cleaned performance data for server consolidation;

wherein assessing the quality of the performance data and cleaning the performance data includes,

evaluating a data structure of each of the input traces by checking metadata of each of the input traces to ensure that the predetermined performance metrics are found in each of the input traces;

excluding any one of the plurality of servers that has an associated one of the input traces with a missing one of the predetermined performance metrics;

applying to each of the input traces one of;

predetermined time stamps, predetermined checks, and predetermined statistical computations;

generating any missing data from each of the input traces that result from applying to each of the input traces;

evaluating a data quality of each of the input traces with at least the any generated missing data; and

checking a time series alignment of each of the input traces by comparing a begin time and an end time of each of the input traces with a predetermined begin time and a predetermined end time of an expected output trace.

18. The method of claim 17, wherein applying to each of the input traces comprises:

determining a number of NULL values for each of the plurality of performance metrics in each of the input traces;

determining a number of data points for each of the plurality of performance metrics in the each input trace that are valid because the valid data points have values within a predetermined range;

determining a number of data values for each of the plurality of performance metrics in the each input trace that are invalid because the invalid data points have values outside the predetermined range;

for each of the invalid data points determined, generating an error message to report the each invalid data point; and

for each of the invalid data points determined, assigning a predetermined value to the each data point.

19. The method of claim 18, wherein any missing data from each of the input traces comprises:

a) calculating a time difference between two adjacent ones of the predetermined time intervals for each of the input traces;

b) determining whether the time difference is greater than a predetermined time difference between time intervals in an expected output trace;

c) upon a determination that the time difference is greater than the predetermined time difference, calculating a number of the predetermined time intervals with missing data for the each input trace;

d) generating an error message indicating the number of the predetermined time intervals with missing data for the each input trace, a begin time for the number, and an end time for the number;

e) generating new data for the missing data for the each performance metric based on predetermined criteria;

f) calculating a missing-row number as a number of groups of consecutive ones of the predetermined time intervals with missing data for the each performance metric;

g) repeating a through f through a last one of the predetermined time intervals in the each input trace; and

h) checking whether there is missing data for the each performance metric at a beginning or an end of the each input trace.

20. The method of claim 19, wherein evaluating a data structure of each of the input traces comprises:

calculating a ratio of the number of NULL values to the number of valid data points for each of the plurality of performance metrics in the each input trace;

determining whether the calculated ratio is greater than a predetermined ratio value and the each performance metric is required; and

upon a determination that the calculated ratio is greater than the predetermined ratio value and the each performance metric is required, generating an error message and marking the server from which the each input trace is collected with an Unchanged status.