Capacity and load analysis in a datacenter

• US 8,738,333 B1
• Filed: 05/25/2010
• Issued: 05/27/2014
• Est. Priority Date: 05/25/2010
• Status: Active Grant

• Current Assignee: Vmware LLC (Broadcom, Inc.)
• Original Assignee: VMware, Inc. (Broadcom, Inc.)
• Inventors: Behera, Somik, McBride, Samuel P., Smith, Justin Daniel
• Primary Examiner(s): Thangavelu, Kandasamy
• Assistant Examiner(s): OCHOA, JUAN CARLOS
• Application Number:
  US12/786,918
• Time in Patent Office: 1,463 Days
• Field of Search: 703/2, 709/224
• US Class Current: 703/2
• CPC Class Codes: G06F 9/5061 Partitioning or combining of resources

Abstract

One or more embodiments provide methods of creating datacenter capacity and/or load models for a datacenter in a virtualized environment; methods of simulating the capacity and/or load models (“What-if” simulations); and methods of visualizing simulation results for assessing the impact of the capacity and/or load models on the datacenter capacity.

Claims

1. A method of simulating a capacity and/or load model in a virtualized environment for a capacity container, which method comprises:

creating a capacity and/or load model;

collecting existing server and virtual machine (VM) attributes for the capacity container, which attributes include measures of capacity utilization;

providing a first and second set of historical data for each of a user-selected number of user-selected time intervals prior to and ending at a predetermined time wherein;

(a) data for the first set in each interval comprises a measure of total capacity in the capacity container in units which are a measure of the number of VMs that can be deployed in the capacity container, wherein the measure of the number of VMs that can be deployed is the number of average virtual machine (AVM) units that can be deployed in the capacity container, and wherein the number of AVM units that can be deployed is determined by dividing a total available CPU demand capacity by a CPU demand of the AVM units and dividing a total available memory capacity by a memory consumption of the AVM units, and utilizing the smaller of the two divisions, and

(b) data for the second set in each interval comprises a measure of total capacity utilization in the capacity container in units which are a measure of the number of VMs that cause the capacity utilization in the capacity container, wherein the measure of the number of VMs that cause utilization is the number of AVM units that cause the utilization in the capacity container, and wherein the number of AVM units that cause the utilization is determined by dividing total CPU demand by the CPU demand of the AVM units, dividing total memory consumption by the memory consumption of the AVM units, and utilizing the smaller of the two divisions;

fitting the first and second sets of historical data to provide a first and a second trend curve; and

visualizing the first and second trend curves.

2. The method of claim 1 which further comprises:

providing a first and second set of data for each of a user-selected number of user-selected time intervals prior to and ending at a predetermined time wherein;

(a) data for the first set in each interval comprises a measure of total capacity in the capacity container, as modified by the capacity and/or load model, in units which are a measure of the number of VMs that can be deployed in the capacity container, and

(b) data for the second set in each interval comprises a measure of the number of VMs that cause capacity utilization in the capacity container, as modified by the capacity and/or load model;

fitting the first and second sets of data to provide a first and a second forecast curve; and

visualizing the first and second forecast curves.

3. The method of claim 1 wherein the number of AVM units that can be deployed is determined by dividing the total available CPU demand capacity by the CPU demand of the AVM units or by dividing a total available memory capacity by the memory consumption of the AVM units;

and

the number of AVM units that cause the utilization is determined by dividing the total CPU demand by the CPU demand of the AVM units or by dividing the total memory consumption by the memory consumption of the AVM units.

4. The method of claim 2 wherein the measure of the number of VMs that can be deployed is the number of average virtual machine (AVM) units that can be deployed in the capacity container;

and

the measure of the number of VMs that cause utilization is the number of AVM units that cause the utilization in the capacity container.

5. The method of claim 4 wherein the number of AVM units that can be deployed is determined by dividing the total available CPU demand capacity by the CPU demand of the AVM units and dividing a total available memory capacity by the memory consumption of the AVM units, and utilizing the smaller of the two divisions;

and

the number of AVM units that cause the utilization is determined by dividing the total CPU demand by the CPU demand of the AVM units, dividing the total memory consumption by the memory consumption of the AVM units, and utilizing the smaller of the two divisions.

6. The method of claim 4 wherein the number of AVM units that can be deployed is determined by dividing the total available CPU demand capacity by the CPU demand of the AVM units or by dividing a total available memory capacity by the memory consumption of the AVM units;

and

the number of AVM units that cause the utilization is determined by dividing the total CPU demand by the CPU demand of the AVM units or by dividing the total memory consumption by the memory consumption of the AVM units.

7. The method of claim 1 wherein a capacity container is one of a datacenter, a server, a cluster of servers, or a virtualized resource pool.

8. The method of claim 1 wherein at least one of the time intervals is one of a day, a week, a month and three months.

9. The method of claim 1 wherein fitting includes using one of linear regression analysis, polynomial regression analysis, exponential regression analysis, logarithmic regression analysis, and power regression analysis.

10. The method of claim 1 wherein the data for the first set and the second set for each time interval is an average of the data over the time interval.

11. The method of claim 1 wherein the data for the first set and the second set for each time interval is a last measured value for the time interval.

12. The method of claim 2 wherein capacity container is one of a datacenter, a server, a cluster of servers, or a virtualized resource pool.

13. The method of claim 2 wherein the time interval is one of a day, a week, a month and three months.

14. The method of claim 2 wherein fitting includes using one of linear regression analysis, polynomial regression analysis, exponential regression analysis, logarithmic regression analysis, and power regression analysis.

15. The method of claim 2 wherein the data for the first set and the second set for each time interval is an average of the data over the time interval.

16. The method of claim 2 wherein the data for the first set and the second set for each time interval is a last measured value for the time interval.

17. The method of claim 1 wherein providing the first and second set of historical data comprises utilizing VMs enabled with a Fault Tolerant feature.

18. The method of claim 1 wherein providing the first and second set of historical data comprises utilizing VMs enabled with a High Availability feature.

19. The method of claim 1 wherein providing the first and second set of historical data comprises utilizing a reserve buffer.

20. The method of claim 2 wherein providing the first and second set of data comprises utilizing VMs enabled with a Fault Tolerant feature.

21. The method of claim 2 wherein providing the first and second set of data comprises utilizing VMs enabled with a High Availability feature.

22. The method of claim 2 wherein providing the first and second set of data comprises utilizing a reserve buffer.