Computing infrastructure resource-workload management methods and apparatuses

• US 10,666,570 B2
• Filed: 11/28/2016
• Issued: 05/26/2020
• Est. Priority Date: 11/28/2016
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Loomba, Radhika, Metsch, Thijs, Katsaros, Grigorios, Butler, Joseph, Mullery, Kevin
• Primary Examiner(s): Duong, Oanh
• Application Number:
  US15/361,905
• Publication Number:
  US 20180152390A1
• Time in Patent Office: 1,275 Days
• CPC Class Codes: H04L 47/70 Admission control; Resource allocation
• CPC Class Codes: H04L 67/1031 Controlling of the operation of servers by a load balancer, e.g. adding or removing servers that serve requests

Abstract

Apparatuses and methods associated with management of computing infrastructure resources and workloads are disclosed herein. In embodiments, an apparatus may comprise a resource-workload manager having: a placement solution generator to generate a plurality of potential resource placement solutions for allocation of various resources of the computing infrastructure to various workloads; one or more utility function calculators to calculate one or more values for one or more provider-centric attributes, and one or more values for one or more customer-centric attributes, for the plurality of potential solutions; an analyzer to analyze the attributes and select one of the potential resource placement solutions; and a resource allocator to allocate the resources, based at least in part on the selected resource placement solution. Other embodiments may be described and/or claimed.

Claims

1. An apparatus for managing resources and their assigned workloads of a computing infrastructure, comprising:

a hardware processor;

a resource-workload manager operated by the hardware processor, the resource-workload manager having;

a placement solution generator to receive resource data and workload data for the computing infrastructure, and generate a plurality of potential resource placement solutions for allocation of the resources of the computing infrastructure to various components of the workloads;

one or more utility function calculators communicatively coupled with the placement solution generator to calculate one or more values for one or more provider-centric attributes, and one or more values for one or more customer-centric attributes, for the plurality of potential resource placement solutions;

an analyzer communicatively coupled to the one or more utility function calculators to analyze the values of the provider-centric and customer-centric attributes and select one of the potential resource placement solutions based at least in part on the analyzed values; and

a resource allocator communicatively coupled to the analyzer to allocate the resources of the computing infrastructure, based at least in part on the selected resource placement solution.

2. The apparatus of claim 1, wherein the one or more utility function calculators include a utility function calculator to calculate the one or more values for the one or more provider-centric attributes, for the plurality of potential resource placement solutions.

3. The apparatus of claim 2, wherein the one or more provider-centric attributes comprise one or more of a revenue attribute, an expenditure attribute, a profit attribute (P), a data locality attribute (), and a data replication attribute (R).

4. The apparatus of claim 3, wherein the plurality of provider-centric attributes comprise a revenue attribute that details an amount of credit ($) a provider of the computing infrastructure would gain by hosting a service with a particular resource placement solution, an expenditure attribute that details a total cost of ownership in credits ($) of the resources in a particular resource placement solution, or a profit attribute (P) of a particular resource placement solution derived from the revenue and expenditure attributes of the particular resource placement solution.

5. The apparatus of claim 3, wherein the plurality of provider-centric attributes comprise a data locality attribute () that details distance or access frequency between compute resources and storage resources of the computing infrastructure, a particular resource placement solution, or a data replication attribute (R) that details a number of possible data/storage replicas the computing infrastructure has to deal with for fault-tolerance/high availability/performance issues for a particular resource placement solution.

6. The apparatus of claim 3 wherein the utility function calculator is to further calculate a provider utility for each of the plurality of potential resource placement solutions, based at least in part on the corresponding calculated values for the plurality of provider-centric attributes for the particular potential resource placement solution:

=(α

₁*(P)+β

₁)*(α

₂*()+β

₂)*(α

₃*(R)+β

₃)

where α

₁, α

₂, and α

₃ are relative weights of profit attribute (P), data locality attribute (), and data replication attribute (R), and β

₁, β

₂, and β

₃ are additive weights that store dependence on other attributes.

7. The apparatus of claim 1, wherein the one or more utility function calculators include a utility function calculator to calculate the one or more values for one or more customer-centric attributes, for the plurality of potential solutions.

8. The apparatus of claim 7, wherein the one or more customer-centric attributes comprise at least two attributes selected from an attribute group that includes a service component locality or anti-locality attribute (L), an availability attribute (A), a network throughput attribute (T), a response time attribute (Rt), and a latency attribute (Lt).

9. The apparatus of claim 8, wherein the one or more customer-centric attributes comprise a service component locality or anti-locality attribute (L) that details requirements of a workload to be hosted by a resource placement solution in one or in separated locations of the computing infrastructure.

10. The apparatus of claim 8, wherein the one or more customer-centric attributes comprise a service component availability attribute (A) that details a desired availability of a service for a particular workload, or a network throughput attribute (T) that detail a desired minimum network throughput that a service of a particular workload needs.

11. The apparatus of claim 8, wherein the one or more customer-centric attributes comprise a service component response time attribute (Rt) that details a time at which a task of a workload is to be completed, and a latency attribute (Lt) that details a maximum desired latency between two service components of a particular workload.

12. The apparatus of claim 8, wherein the utility function calculator is to further calculate a customer utility _e for each of the plurality of potential resource placement solutions, based at least in part on the corresponding calculated values for the one or more provider-centric attributes for the particular potential resource placement solution:

_e=(α

₄*(L)+β

₄)*(α

₅*(A)+β

₅)*(α

₆*(T)+β

₆)*(α

₇*(Rt)+β

₇)*(α

₈*(Lt)+β

₈)

where α

₄, α

₅, α

₆, α

₇, and α

₈ are relative weights of locality or anti-locality attribute (L), availability attribute (A), network throughput attribute (T), response time attribute (Rt), and latency attribute (Lt), and β

₄, β

₅, β

₆, β

₇, and β

₈ are additive weights that store dependence on other attributes.

13. The apparatus of claim 1, wherein the hardware processor includes one or more processor cores respectively dedicated to operate the one or more utility function calculators.

14. The apparatus of claim 1, wherein the hardware processor includes one or more processor cores, and one or more hardware accelerators, wherein the one or more hardware accelerators are respectively dedicated to operate the one or more utility function calculators.

15. A method for managing resources and their assigned workloads of a computing infrastructure, comprising:

receiving, by a placement solution generator of a resource-workload manager operating on a computing device, resource data and workload data for the computing infrastructure;

generating, by the placement solution generator, a plurality of potential resource placement solutions for allocation of the resources of the computing infrastructure to various components of the workloads;

calculating, by one or more utility function calculators of the resource-workload manager, one or more values for one or more provider-centric attributes, and one or more values for one or more customer-centric attributes, for the plurality of potential resource placement solutions;

analyzing, by an analyzer of the resource-workload manager, the values of the provider-centric and customer-centric attributes and selecting one of the potential resource placement solutions, based at least in part on the analyzed values; and

allocating, by a resource allocator of the resource-workload manager, the resources of the computing infrastructure, based at least in part on the selected resource placement solution.

16. The method of claim 15, wherein calculating comprises calculating the values for the one or more provider-centric attributes, for the plurality of potential resource placement solutions;

wherein the one or more provider-centric attributes comprise one or more of a revenue attribute, an expenditure attribute, a profit attribute (P), a data locality attribute (D), and a data replication attribute (R); and

wherein calculating comprises calculating values for one or more customer-centric attributes, for the plurality of potential solutions;

wherein the one or more customer-centric attributes comprise at least two attributes selected from an attribute group that includes a service component locality or anti-locality attribute (L), an availability attribute (A), a network throughput attribute (T), a response time attribute (Rt), and a latency attribute (Lt).

17. The method of claim 16, wherein calculating further comprises calculating a provider utility _s for each of the plurality of potential resource placement solutions, based at least in part on the corresponding calculated values for the plurality of provider-centric attributes for the particular potential resource placement solution:

_s=(α

₁*(P)+β

₁)*(α

₂*(R)+β

₂)*(α

₃*(R)+β

₃)

where α

₁, α

₂, and α

₃ are relative weights of profit attribute (P), data locality attribute (D), and data replication attribute (R), and β

₁, β

₂, and β

₃ are additive weights that store dependence on other attributes.

18. The method of claim 17, wherein calculating further comprises calculating a customer utility _e for each of the plurality of potential resource placement solutions, based at least in part on the corresponding calculated values for the one or more provider-centric attributes for the particular potential resource placement solution:

₆=(α

₄*(L)+β

₄)*(α

₅*(A)+β

₅)*(α

₆*(T)+β

₆)*(α

₇*(Rt)+β

₇)*(α

₈*(Lt)+β

₈)

where α

₄, α

₅, α

₆, α

₇, and α

₈ are relative weights of locality or anti-locality attribute (L), availability attribute (A), network throughput attribute (T), response time attribute (Rt), and latency attribute (Lt), and β

₄, β

₅, β

₆, β

₇, and β

₈ are additive weights that store dependence on other attributes.

19. One or more non-transitory computer-readable media (CRM) comprising instructions that cause a computing device, in response to execution of the instructions by a hardware processor, to operate a resource-workload manager of a computing infrastructure to:

receive resource data and workload data for the computing infrastructure;

generate a plurality of potential resource placement solutions for allocation of the resources of the computing infrastructure to various components of the workloads;

calculate one or more values for one or more provider-centric attributes, and one or more values for one or more customer-centric attributes, for the plurality of potential resource placement solutions;

analyze the values of the provider-centric and customer-centric attributes and select one of the potential resource placement solutions, based at least in part on the analyzed values; and

allocate the resources of the computing infrastructure, based at least in part on the selected resource placement solution.

20. The non-transitory CRM of claim 19, wherein to calculate comprises to calculate the values for the one or more provider-centric attributes, for the plurality of potential resource placement solutions.

21. The non-transitory CRM of claim 20, wherein the one or more provider-centric attributes comprise one or more of a revenue attribute, an expenditure attribute, a profit attribute (P), a data locality attribute (D), and a data replication attribute (R).

22. The non-transitory CRM of claim 21, wherein to calculate further comprises to calculate a provider utility _s for each of the plurality of potential resource placement solutions, based at least in part on the corresponding calculated values for the plurality of provider-centric attributes for the particular potential resource placement solution:

_s=(α

₁*(P)+β

₁)*(α

₂*(D)+β

₂)*(α

₃*(R)+β

₃)

where α

₁, α

₂, and α

₃ are relative weights of profit attribute (P), data locality attribute (D), and data replication attribute (R), and β

₁, β

₂, and β

₃ are additive weights that store dependence on other attributes.

23. The non-transitory CRM of claim 19, wherein one or more utility function calculators include a utility function calculator to calculate values for one or more customer-centric attributes, for the plurality of potential solutions;

wherein the one or more customer-centric attributes comprise at least two attributes selected from an attribute group that includes a service components locality or anti-locality attribute (L), an availability attribute (A), a network throughput attribute (T), a response time attribute (Rt), and a latency attribute (Lt).

24. The non-transitory CRM of claim 23, wherein to calculate further comprises to calculate a customer utility _e for each of the plurality of potential resource placement solutions, based at least in part on the corresponding calculated values for the one or more provider-centric attributes for the particular potential resource placement solution:

_e=(α

₄*(L)+β

₄)*(α

₅*(A)+β

₅)*(α

₆*(T)+β

₆)*(α

₇*(Rt)+β

₇)*(α

₈*(Lt)+β

₈)

where α

₄, α

₅, α

₆, α

₇, and α

₈ are relative weights of locality or anti-locality attribute (L), availability attribute (A), network throughput attribute (T), response time attribute (Rt), and latency attribute (Lt), and β

₄, β

₅, β

₆, β

₇, and β

₈ are additive weights that store dependence on other attributes.