Method and apparatus for improving efficiency of a cloud platform based on node usage

1. A method for provisioning a cloud platform, performed by a cloud mediator interface which is arranged on each cloud node and performs a role of a bridge between a cloud system and a Ganglia, wherein each cloud node comprises a multi core CPU, one or more network I/O units, one or more disk I/O units, and one or more non-transitory memories, the method comprising:

collecting usage data of a cloud node over intervals of time;

generating and storing an energy usage profile for each cloud node by using the collected usage data;

acquiring real-time usage data from the cloud node;

comparing the real-time usage data with the generated energy usage profile so as to calculate a recommendation value; and

provisioning the cloud platform based on the recommendation value,

wherein each of the storing and the acquiring comprises calculating variance between stress and idle levels of nodes, and

wherein when a new node determined to have a similar specification to any profiled node is added to a cloud, the generating comprises immediately adding the new node without an additional profiling process, for avoiding profiling nodes with similar hardware specifications to save evaluation time.

2. The method of claim 1 , wherein the usage data includes any one or any combination of any two or more of CPU usage, network input/output (I/O), disk I/O, and usage memory data.

3. The method of claim 1 , wherein the energy usage profile is tagged with information comprising any one or any combination of any two or more of a CPU type, available cores, available memory, and hypervisor type.

4. The method of claim 3 , wherein each of the storing and the acquiring further comprises

mapping an actual usage to a variance scale so as to calculate load to power ratio,

calculating a total power by adding power consumed at hardware components, and

calculating a consumed power of a task by multiplying an amount of running time for the task by a total power of the task.

5. The method of claim 4 , wherein the hardware components comprise any one or any combination of any two or more of a memory, storage drives, fans, power supply units, and power usage of the hardware components provided in a CPU specification.

6. The method of claim 1 , wherein the generating and storing of the energy usage profile comprises storing a most energy-efficient usage profile.

7. The method of claim 1 , wherein the generating and storing of the energy usage profile comprises storing usage profiles in a form that is classified according to an application, a virtual machine, or node attributes.

8. The method of claim 1 , wherein the generating and storing of the energy usage profile comprises extracting usage patterns from a total usage.

9. The method of claim 8 , wherein the generating and storing of the energy usage profile further comprises matching real-time values of usage parameters to profile parameters stored in the usage patterns based on a threshold.

10. The method of claim 9 , wherein

the generating and storing of the energy usage profile further comprises matching the real-time values of the usage parameters to profiles, and

in response to detecting a possible match, a profile among the profiles is returned.

11. The method of claim 9 , wherein the generating and storing of the energy usage profile further comprises returning an energy efficiency scheme for a particular usage as an intermediate recommendation value, in response to detecting an exact match.

12. The method of claim 11 , wherein the recommendation value is calculated by adding a plurality of intermediate recommendation values, including the intermediate recommendation value, to a plurality of recommendation values.

13. The method of claim 1 , further comprising directly adding a another new node by limiting a profiling process of the another new node, in response to determining that a hardware specification of the another new node is similar to a hardware specification of the cloud node.

14. An apparatus for provisioning a cloud platform, wherein the apparatus is arranged on each cloud node and performs a role of a bridge between a cloud system and a Ganglia, and wherein each cloud node comprises a multi core CPU, one or more network I/O units, one or more disk I/O units, and one or more non-transitory memories, the apparatus comprising:

a profiler configured to collect usage data of a cloud node over intervals of time, and generate and store an energy usage profile for each cloud node by using the collected usage data;

a real-time usage feedback receiver configured to acquire real-time usage data from the cloud node; and

a recommendation engine configured to compare the real-time usage data with the generated energy usage profile so as to calculate a recommendation value,

wherein the cloud platform is provisioned based on the recommendation value,

wherein either one or both of the profiler and the real-time usage feedback receiver is configured to calculate variance between stress and idle levels of nodes, and

wherein when a new node determined to have a similar specification to any profiled node is added to the cloud, the profiler immediately adds the new node without an additional profiling process for avoiding profiling nodes with similar hardware specifications to save evaluation time.

15. The apparatus of claim 14 , wherein the usage data includes any one or any combination of any two or more of CPU usage, network input/output (I/O), disk I/O, and usage memory data.

16. The apparatus of claim 14 , wherein the energy usage profile is tagged with information comprising any one or any combination of any two or more of a CPU type, available cores, available memory, and hypervisor type.

17. The apparatus of claim 16 , wherein each of the profiler and the real-time usage feedback receiver is configured to map an actual usage to a variance scale so as to calculate load to power ratio, calculate a total power by adding power consumed at hardware components, and calculate a consumed power of a task by multiplying an amount of running time for the task by a total power of the task.

18. The apparatus of claim 14 , wherein the profiler is further configured to store a most energy-efficient usage profile.

19. The apparatus of claim 14 , wherein the profiler is further configured to store usage profiles in a form that is classified according to an application, a virtual machine, or node attributes.