Virtual Computing Services Deployment Network

• US 20130212282A1
• Filed: 12/17/2012
• Published: 08/15/2013
• Est. Priority Date: 10/20/2006
• Status: Active Grant

• Current Assignee: Vmware LLC (Broadcom, Inc.)
• Original Assignee: Desktone Incorporated (Broadcom, Inc.)
• Inventors: Battersby, Clinton B., Gaffney, Paul J., Pulier, Eric Y.
• Granted Patent: US 10,110,512 B2
• Time in Patent Office: Days
• US Class Current: 709/226
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: G06Q 10/0631 Resource planning, allocation, distributing or scheduling for enterprises or organisations
• CPC Class Codes: H04L 47/822 Collecting or measuring resource availability data

Abstract

A business method associates provisioning costs with a usage history indicative of user computing demand, and coalesces the cost data to identify an appropriate provisioning level balancing the provisioning cost and the usage demand cost. Conventional computing environments suffer from the shortcoming of being prone to overprovisioning or underprovisioning a user. Such misprovisioning is inefficient because it denotes underutilized computing resources or ineffective and/or disgruntled users. Costs increase either due to the excessive hardware bestowed on the overprovisioned user, or in support costs addressing the underprovisioned user. Configurations herein substantially overcome such shortcomings by defining a policy indicative of overprovisioning and underprovisioning indicators (misprovisioning flags), and defining rules to specify a triggering event indicating the need to reassess the provisioning of a user.

Claims

1. A method of managing computing infrastructure costs, comprising:

defining a set of computing classes, each computing class associated with a per user cost;

accumulating a usage history corresponding to each user in a set of users, the usage history indicative of computing demands of each user;

analyzing the computing demands to define a user profile for each user, the computing demands indicative of a provisioning cost of the user;

provisioning each user in a particular computing class based on the user profile, the per user cost and the provisioning cost; and

periodically reanalyzing the computing demands to selectively reprovision the user.

2. The method of claim 1 wherein provisioning further comprises:

comparing the user profile to a provisioning policy, the provisioning policy defining a best fit of computing resources to user demands;

determining if the user is misprovisioned based on the comparison, misprovisioning indicative of a user having resources inconsistent with the user demands indicated by the user profile; and

selecting a different computing class for the user if the provisioning policy is indicative of a misprovisioned user.

3. The method of claim 2 wherein the provisioning policy defines a balance between overprovisioned and underprovisioned users, an overprovisioned user associated with a higher operational cost than an underprovisioned user, an underprovisioned user associated with a higher support cost than an overprovisioned user.

4. The method of claim 3 wherein the operational cost encompasses costs of providing the user with computer processing capability, and the support cost encompasses the cost of responding to user inquiries of computer processing capability.

5. The method of claim 4 wherein the overprovisioned user receives a greater share of computing resources than an underprovisioned user, each user associated with a provisioning tradeoff threshold, the tradeoff threshold indicating the lowest operation cost supportable without increasing the support cost.

6. The method of claim 1 further comprising a plurality of computing service deployments, each deployment having a network location and operable to provision at least one class of computing services, including:

identifying a computing service corresponding to each particular class, the computing service provided by one of the plurality of computing services deployments;

associating a fee with each of the computing services in each of the computing services deployments;

evaluating the feasibility of the computing service identified for each user, evaluating including the network location of the computing service location relative to the particular user; and

selectively modifying the computing service of the user based on the evaluation.

7. The method of claim 6 further comprising evaluating the feasibility using a best-fit metric, the best-fit metric defining a cost effective provisioning level for the user, further comprising:

defining a set of rules, the rules specifying overprovisioning and underprovisioning thresholds, wherein the rules balance enterprise cost with user satisfaction to balance overprovisioning and underprovisioning, such that the computed provisioning level is independent of the provisioning level of other users.

8. The method of claim 7 further comprising defining the best fit metric in a policy, the policy having a set of rules, each of the rules identifying a computing performance concern and a resource cost concern, the rule further specifying a tradeoff threshold between the performance concern and the resource cost concern.

9. The method of claim 8 wherein:

the computing performance concerns include allocated processor speed, allocated memory, allocated disk space;

the resource cost concerns include help desk calls, equipment requisitions, and external grid computing requests; and

periodically reanalyzing occurs upon triggering events including threshold specifying at least one of page faults, response time, and CPU utilization

10. The method of claim 2 further comprising defining a plurality of computing classes, each computing class associated with a resource cost and a performance factor, the computing classes comprising at least one of:

a terminal server class operable to support a plurality of users on a common operating system in a shared memory pool;

a hypervisor class providing a dedicated OS and memory allocation to each user; and

a grid class providing high performance computing services.

11. The method of claim 3 wherein the classes include a grid class, further comprising:

identifying internal and external provisioning sources and costs thereof;

computing a feasibility of external provisioning, external provisioning including provisioning in a grid computing class; and

selectively provisioning the user in the grid class based on the computed feasibility.

12. A virtual computing service cluster, comprising:

a set of deployments, each deployment operable to provide at least one of a predefined set of computing classes, each computing class associated with a per user cost;

a profile database accumulating a usage history corresponding to each user in a set of users, the usage history indicative of computing demands of each user;

a provisioner in the deployment for analyzing the computing demands to define a user profile for each user, the computing demands indicative of a provisioning cost of the user; and

an instance manger for provisioning each user in a particular computing class based on the user profile, the per user cost and the provisioning cost, the provisioner periodically reanalyzing the computing demands to selectively reprovision the user.

13. The computing service cluster of claim 12 wherein the provisioner is further operable to:

compare the user profile to a provisioning policy, the provisioning policy defining a best fit of computing resources to user demands;

determine if the user is misprovisioned based on the comparison, misprovisioning indicative of a user having resources inconsistent with the user demands indicated by the user profile; and

select a different computing class for the user if the provisioning policy is indicative of a misprovisioned user.

14. The computing service cluster of claim 13 wherein the provisioning policy defines a balance between overprovisioned and underprovisioned users, an overprovisioned user associated with a higher operational cost than an underprovisioned user, an underprovisioned user associated with a higher support cost than an overprovisioned user.

15. The computing service cluster of claim 14 wherein the operational cost encompasses costs of providing the user with computer processing capability, and the support cost encompasses the cost of responding to user inquiries of computer processing capability, such that the overprovisioned user receives a greater share of computing resources than an underprovisioned user, each user associated with a provisioning tradeoff threshold, the tradeoff threshold indicating the lowest operation cost supportable without increasing the support cost.

16. The computing service cluster of claim 12 further comprising a plurality of computing service deployments, each deployment having a network location and operable to provision at least one class of computing services, further comprising best-fit logic operable to identify a computing service corresponding to each particular class, the computing service provided by one of the plurality of computing services deployments, the best-fit logic associating a fee with each of the computing services in each of the computing services deployments;

the best-fit logic further evaluating the feasibility of the computing service identified for each user, evaluating including the network location of the computing service location relative to the particular user; and

the provisioner operable to selectively modify the computing service of the user based on the evaluation.

17. The computing service cluster of claim 16 wherein the best-fit logic defines a cost effective provisioning level for the user, further comprising:

defining a set of rules, the rules specifying overprovisioning and underprovisioning thresholds, wherein the rules balance enterprise cost with user satisfaction to balance overprovisioning and underprovisioning, such that the computed provisioning level is independent of the provisioning level of other users, the best fit logic defined in response to a policy, the policy having a set of rules, each of the rules identifying a computing performance concern and a resource cost concern, the rule further specifying a tradeoff threshold between the performance concern and the resource cost concern.

18. The computing service cluster of claim 12 further comprising a plurality of computing classes, each computing class associated with a resource cost and a performance factor, the computing classes comprising:

a terminal server class operable to support a plurality of users on a common operating system in a shared memory pool;

a hypervisor class providing a dedicated OS and memory allocation to each user; and

a grid class providing high performance computing resources substantially greater than the hypervisor class.

19. The computing service cluster of claim 18 wherein the computing classes define the grid class as an external provisioning source, the provisioner further invoking the best fit logic for:

identifying internal and external provisioning sources and costs thereof;

computing a feasibility of external provisioning, external provisioning including provisioning in a grid computing class; and

selectively provisioning the user in the grid class based on the computed feasibility.

20. A business method of virtual desktop computing comprising:

identifying a service corresponding to a provisioning level;

associating a fee with the service level;

identifying internal and external provisioning sources and costs thereof; and

computing a feasibility of external provisioning based on a best-fit logic, the a best fit logic based on a policy, the policy defining rules selectable by an enterprise for provisioning users, further comprising;

selectively internally provisioning a user by;

instantiating a computing environment to support the provisioned environment on the server;

analyzing further comprises applying a set of rules in a usage policy to the user provide to compute a provisioning class;

identifying a provisioning deployment server nearest to the user device supporting the computing environment;

or

selectively externally provisioning a user by provisioning the user in the grid class based on the computed feasibility, the grid class operable for specialized instances of high-performance computing substantially greater than that available via internal provisioning.

21. The method of claim 20 wherein the virtual desktop computing further comprises computing via a grid, the grid having an interconnected collection of computing resources not owned by an invoker or the grid but available on a demand basis as a service, the service defining a variable level of commitment and flexibility upon which the service may be invoked.