TECHNICAL FIELD

The present disclosure relates generally to computer system management and more particularly to a system and method for dynamic resource provisioning for job placement.

BACKGROUND

Reliability and speed of processing activities has become increasingly important, particularly in large groups of computer systems where multiple users may require system resources to perform various processing activities. These processing activities may compete with one another for system resources. Overloading a server in a computer network with too many processing activities may cause the server to operate poorly or crash. Accordingly, the ability to distribute processing activities intelligently within a computer system, for example a set of servers, may increase overall operational efficiency.

SUMMARY

The present disclosure relates generally to a system and method for dynamic resource provisioning for job placement. In particular embodiments, a method for dynamic resource provisioning for job placement includes receiving a request to perform a job on an unspecified computer device. One or more job criteria for performing the job and one or more utilization criteria for performing the job are determined. The one or more job criteria define one or more operational characteristics needed for a computer device to perform the job. The method further includes providing a list of available computer devices. The list of available computer devices includes a plurality of computer devices currently provisioned to perform computer operations. From the list of available computer devices, a list of suitable computer devices for performing the job is determined by comparing operational characteristics for each available computer device with the job criteria. The list of suitable computer devices includes one or more computer devices having operational characteristics that satisfy the job criteria. The method further includes using the utilization criteria to determine whether one or more underutilized computer devices exist on the list of suitable computer devices wherein the one or more underutilized computer devices have a suitable level of utilization for performing the job. If the one or more underutilized computer devices exist, the job is forwarded to one of the one or more underutilized computer devices.

Technical advantages of particular embodiments of the present disclosure may include providing a system and method for evaluation of real-time, aggregated overall utilization of computer devices, according to user-definable criteria for computer device allocation, and computer device provisioning. Flexible, user-definable criteria (e.g., job criteria) and rules for evaluating the overall utilization of individual computer devices may be applied to user defined resource pools to select the best computer device available for performing a job. Moreover, the evaluation criteria (e.g., job criteria) are varied and therefore may provide sufficient information to make real-time intelligent decisions about the state of user defined resource pools. As an additional technical advantage of the present disclosure, user-defined criteria may be used to determine the type of new computer device to be provisioned for a new job if no suitable computer device is identified within a pool of active computer devices. Moreover, the ability to factor in hardware and software component information (e.g., as job criteria) in addition to a variety of utilization metrics may provide a further technical advantage in determining the â€œbestâ€ location where work (e.g., a job) may be placed. Yet another technical advantage of the present disclosure is the ability to segregate server level rules from service level resource utilization.

Other technical advantages of the present disclosure will be readily apparent to one skilled in the art from the following figures, descriptions, and claims. Moreover, while specific advantages have been enumerated above, various embodiments may include all, some, or none of the enumerated advantages.

BRIEF DESCRIPTION OF THE DRAWINGS

For a more complete understanding of the present disclosure and its advantages, reference is now made to the following descriptions, taken in conjunction with the accompanying drawings, in which:

FIG. 1 illustrates an example system for dynamic resource provisioning for job placement according to an example embodiment of the present disclosure; and

FIG. 2 illustrates an example method for dynamic resource provisioning for job placement according to an example embodiment of the present disclosure.

DETAILED DESCRIPTION OF EXAMPLE EMBODIMENTS

When a user of a computer system submits a job (e.g., a computing operation) for execution, the computer system may have several parallel computer devices (e.g., servers) available to perform the job. However, not every one of the available computer devices may be appropriate or even suitable for performing the job. For example, some of the available computer devices may not be adequately provisioned to perform the job while others may be too busy to perform the job in spite of being adequately provisioned.

In cases where multiple computer devices are available to perform a job, it may be desirable to consider the appropriateness of each available computer device for performing the job prior to assigning the job so that the best computer device for performing the job may be identified. This process may be especially beneficial in the context of data centers that may have hundreds or even thousands of computer devices from which to choose. To identify the best computer device available for a job, the computer system may monitor the available computer devices on an approximately real-time basis and divert each new job to the most appropriate computer device as each new job comes in. This process of dynamically identifying the best available computer device for performing a job at any given moment may help to balance workload across the computer system, likely increasing reliability and efficiency of the plurality of computer devices.

In particular embodiments, the computer system may use an overall utilization value (e.g., a calculated aggregate of performance metrics that may be based on user-defined parameters), individually collected utilization metrics, events, and user-defined parameters to determine the best computer device available for a particular job. Rule statements (or basic evaluations) may be put together to form complex decision making rules and may be used to evaluate some or all of these criteria, for example, to identify the best computer device available to perform a job, to make decisions about when it is appropriate to take user defined actions, or to provide a new computer device for new work to be performed. The rules may be evaluated on an individual server basis (e.g., for a single computer device) or on a group of computer devices (e.g., a collection of servers). In some cases it may be appropriate to look at the individual computer device as opposed to a group of computer devices because an individual computer device may be critical to a business function and may experience conditions or utilization which indicate that it is over-utilized while the remaining computer devices in the group are not being utilized or are being under utilized. The opposite may be true: the group of computer devices may be heavily utilized or conditions may exist for the service which indicates that the whole group is over-utilized even though the individual computer devices in the group are only moderately utilized.

In some cases, depending on the user-defined parameters and what computer devices are currently available, there may be no appropriate destinations where the work can be placed, and therefore, the computer system may provision a new computer device that will satisfy the user-defined criteria and will be available to perform the new work. For example, if a particular action or job requires a system with a minimum of 2 Giga Bytes (â€œGBâ€ ) of RAM and a 2.4 Giga hertz (â€œGHzâ€ ) CPU and requires these two metrics to be less than 10% utilized, the computer system may provision a new computer device if none are currently available that fit the given criteria.

The computer system may have the ability to monitor and aggregate resource utilization at an individual computer device level or group level. This, in turn, may enable the computer system to act as an intelligent, real-time resource broker to applications that may require this capability. In particular embodiments, through the use of an aggregated resource utilization value, the computer system may make intelligent decisions based on the user-defined policies, requirements, and real-time analysis of the available computer devices.

FIG. 1 illustrates an example system 100 for dynamic resource provisioning for job placement according to an example embodiment of the present disclosure. System 100 generally includes a client machine 102 operable to submit a job request to a brokering engine 106 residing on a server machine 108, the job request including a request to for brokering engine 106 to find a most suitable computer device 112 available for processing a job 104. Brokering engine 106 may monitor an active pool 110 of computer devices 112 and identify the best computer device 112 available in active pool 110 for performing job 104 based on one or more job criteria 105 (e.g., particular hardware requirements) and utilization criteria 107 (e.g., particular utilization requirements) needed to perform job 104. In particular embodiments job criteria 105 and utilization criteria 107 may be submitted to brokering engine 106 as part of the request for a best computer device 112 from client machine 102.

To identify the most suitable computer device 112 for job 104, brokering engine may compare the operational characteristics of each computer device 112 against job criteria 105 to determine a list of computer devices having suitable operational characteristics, and may compare the usage statistics 114 from each computer device 112 on the list against utilization criteria 107 to determine which computer device(s) 112 have a suitable level of activity for performing job 104. Once brokering engine 106 has determined a list of devices 112 having suitable operational characteristics and a suitable level of utilization for performing job 104, brokering engine may return the list of suitable computer devices 112 to client machine 102. and client machine 102 may forward the job to one of the devices on the list for execution. In particular embodiments, if brokering engine 106 determines that none of the computer devices 112 in active pool 110 are appropriate for performing job 104, for example, if none of computer devices 112 are properly provisioned, or if all of computer devices 112 are too busy to perform job 104, brokering engine 106 may provision a new computer device 112 from an inactive pool 118 of computer devices 112 to perform job 104.

The components of system 100 (e.g., client machine 102, server machine 108, and computer devices 112) may be coupled to each other through a network 120 via one or more links 116. In particular embodiments, links 116 may each include one or more wireline, wireless, or optical links. In particular embodiments, one or more links 116 each include a local area network (â€œLANâ€ ), a wireless local area network (â€œWLANâ€ ), a wide area network (â€œWANâ€ ), a metropolitan area network (â€œMANâ€ ), a portion of the Internet, or another link 116 or a combination of two or more such links 116. Each of the components coupled to network 120 communicate with each other via use of network 120.

In particular embodiments, network 120 may include one or more LANs, WLANs, WANs, MANs, a portion of the Internet, or another form of network or a combination of two or more such networks. The present disclosure contemplates any suitable network 120 or combination of networks 120. In particular embodiments, components of system 100 may be distributed across multiple cities or geographical regions connected by network 120. In particular embodiments, network 120 may be represented by multiple distinct, but interconnected networks that share components or distinctly contain similar components.

Once interconnected, the components of system 100 may work in conjunction with one another to provide computing services to an individual, company, or other organizational entity. One of ordinary skill in the art will appreciate that the above-described architecture of system 100 was presented for the sake of explanatory simplicity and will further appreciate that the components of system 100 may vary in number and arrangement without departing from the scope of the present disclosure.

In a typical scenario, the computer devices 112 in active pool 110 may receive and perform jobs 104 from one or more client machines 102. A job 104 may be any computing-based task submitted by client machine 102 (or other suitable source) to active pool 110 for execution. As an example and not by way of limitation, a job 104 may include an operation or a batch of operations to be performed by a single instance of a computer program residing on one of computer devices 112. For instance, job 104 may be a batch of payroll accounts to be tabulated at the end of a payroll period for employees of a company using the company's payroll program.

Client machine 102 may submit a job 104 to active pool 110 for execution for any number of reasons such as, for example, because client machine 102 lacks the proper software to perform job 104 or because client machine 102 does not have enough processing power to complete the job in a timely manner. In any case, active pool 110 may serve as repository of available computer devices 112 that are currently provisioned and available to perform computing-based operations (e.g., jobs 104) on behalf of client machines 102. For example, active pool 110 may consist of a collection of servers.

In particular embodiments, system 100 may further include an inactive pool 118 of available computer devices 112 that are not currently provisioned to perform computing-based operations on behalf of client machines 102. For example, the computer devices 112 in inactive pool 118 may not be provisioned to perform the job 104 because they are turned off, are offline, or are simply held in reserve unless needed. In any case, the dividing line between the computer devices 112 in active pool 110 and the computer devices 112 in inactive pool 118 may be defined in certain circumstances by whether the computer device 112 is readily able to perform computing-based operations.

In particular embodiments, the distinction between whether a computer device 112 is included in active pool 110 or inactive pool 118 may be indicated by one or more inherent properties of the computer device 112. For example, if a computer device 112 is turned off it may be included in inactive pool 118. As another example and not by way of limitation, if a computer device 112 lacks an operating system it may be included in inactive pool 118. In particular embodiments, a user of system 100 may designate whether a computer device 112 is included in active pool 110 or inactive pool 118 using, for example, an asset profile which includes information regarding computer device 112. One of ordinary skill in the art will appreciate that the above-described criteria for defining active pool 110 and inactive pool 118 were presented for the sake of explanatory simplicity and will further appreciate that the present disclosure contemplates using any suitable method or criteria to define an active pool 110 of computer devices 112 and an inactive pool 118 of computer devices 112.

In particular embodiments, a job 104 may be submitted to active pool 110 from one or more client machines 102. A client machine 102 may be any component of hardware or software or combination of two or more such components operable to submit a job 104 to a computer device 112 for execution. As an example and not by way of limitation, client machine 102 may be considered a special type of computer device 112 distinguished by the fact that it acts as the source of job 104. Typically, client machine 102 resides outside of active pool 110; however, in particular embodiments, client machine 102 may be among the computer devices 112 included in active pool 110.

Client machine 102 may include an interface 120, one or more processors 122, and a memory 124 comprising or capable of receiving logical program instructions recorded on a tangible computer readable media 126 (e.g., a cd-rom, a flash drive, a floppy disk, etc.). In particular embodiments, a user may interact with a client machine 102 either directly or indirectly via any suitable input device(s) (e.g., a keyboard or a mouse) and output device(s) (e.g., a monitor) connected either directly or indirectly to client machine 102. Example embodiments of client machine 102 may include a personal computer, a laptop, a personal digital assistant, or other processor-controlled device capable of submitting a job 104 for execution by a computer device 112.

In particular embodiments, when client machine 102 submits a job 104 to active pool 110 for execution, client machine 102 may be indifferent as to the particular computer device 112 that will ultimately perform job 104. Accordingly, in particular embodiments, client machine 102 may use a brokering engine 106 to select the best (e.g., most suitable) computer device 112 available in active pool 110 for performing job 104. Accordingly, client machine 102 may request brokering engine 106 to find the best computer device 112 available in active pool 110 to perform job 104 and may supply one or more job criteria 105 and one or more utilization criteria 107 to brokering engine 106 to assist brokering engine 106 in identifying the best computer device 112 for performing job 104 as described in more detail below.

Brokering engine 106 may be any component of hardware or software or combination of two or more such components capable of applying the job criteria 105 and utilization criteria 107 associated with job 104 with the computer devices 112 in active pool 110 to determine the best computer device 112 available to perform job 104. As an example and not by way of limitation, brokering engine 106 may be a software program that logically resides between client machine 102 and active pool 110. As an additional example and not by way of limitation, brokering engine 106 may be a software program that resides on client machine 102 or on a computer device 112 in active pool 110. In any case, when brokering engine 106 receives a request to find the best computer device 112 to perform job 104 from client machine 102, brokering engine 106 may identify the best computer device 112 available to perform job 104 to client machine 102 and client machine 102 may forward job 104 to the best computer device 112 for execution based on the decision of brokering engine 106.

A computer device 112 may be any processor-controlled device or combination of two or more such devices capable of executing a computing operation (e.g., a job 104). As an example and not by way of limitation, a computer device 112 may be a host computer, workstation, personal computer, web server, file server, or other server. In particular embodiments, a computer device 112 may include an interface 120, one or more processors 122, and a memory 124 comprising or capable of receiving logical program instructions recorded on a tangible computer readable media 126 (e.g., a cd-rom, a flash drive, a floppy disk, etc.). In particular embodiments, a user may interact with a computer device 112 either directly or indirectly via any suitable input device(s) (e.g., a keyboard or a mouse) and output device(s) (e.g., a monitor) connected either directly or indirectly to computer device 112.

Processor 122 may refer to any suitable device operable to execute instructions and manipulate data to perform operations for computer devices 112. Processor 122 may include, for example, any type of central processing unit (CPU). Memory 124 may be any form of volatile or non-volatile memory including, without limitation, magnetic media, optical media, random access memory (RAM), read-only memory (ROM), removable media, or any other suitable local or remote memory component. Interface 120 may comprise any hardware, software, or encoded logic operable to send and receive information to and from other components of system 100 such as client machine 102, server machine 108, or other computer devices 112.

In particular embodiments, a computer device 112 may include any number and type of software applications such as one or more operating systems (e.g., MS-DOS, PC-DOS, MAC-OS, WINDOWS, UNIX, or other suitable operating system), database applications (e.g. MICROSOFT EXCEL, MICROSOFT ACCESS), word processing applications (e.g., MICROSOFT WORD), proprietary applications, or other suitable software applications, one or more of which may be suitable for performing job 104.

When a computer device 112 performs a computer-based operation (e.g., a job 104), various resources of the computer device 112 (e.g., processor capacity, memory capacity, etc.) may be devoted to performing that computer-based operation. Thus, the amount of resources that a computer device 112 has available at a given moment in time may limit the number of jobs 104 that the computer device 112 may perform without reaching its operational capacity (e.g., the point at which the computer device 112 has no further resources available). Moreover, as the usage of a computer device 112 reaches its operational capacity, the performance of computer device 112 may decline. For example, computer device 112 may run slower as the usage of the computer device 112 nears its operational capacity. Furthermore, computer device 112 may crash or exhibit other problems if the usage of computer device 112 exceeds its operational capacity.

The usage of computer device 112 (e.g., the activity level of computer device 112) may be monitored, for example, using one or more utilization metrics 114 gathered from computer device 112. For example, the utilization metrics 114 may be regularly gathered by a collection agent residing on each computer device 112 and delivered to brokering engine 106 at regular intervals (e.g., every 30 seconds). Utilization metrics 114 may include any indicator(s) suitable for indicating the amount of resources that a computer device 112 has available to perform a job 104. For example, a utilization metric 114 may be a single numeric value that represents a particular type and level of resource usage currently taking place at computer device 112. Though utilization metrics 114 may vary depending upon the characteristics of the computer device 112 being monitored, example utilization metrics 114 may include system level metrics (e.g., clustered environment and virtualized environment) as well as applications level metrics. For example, utilization metrics 114 may include memory utilization metrics, such as paging metrics, swapping metrics, and available memory metrics. As another example, utilization metrics 114 may include CPU utilization metrics at the particular computer device 112. For the sake of explanatory clarity, a nonlimiting list of example utilization metrics 114 that may be used to monitor the activity level of a computer device 112 follows at the end of the specification.

In particular embodiments, one or more utilization metrics 114 may be combined into a single overall utilization metric that may generally describe the current activity level of a computer device 112. Moreover, the overall utilization metric (or any other utilization metric 114) for a computer device 112 may be a normalized value relative to other computer devices 112 to enable system 100 to easily compare the usage level of one computer device 112 against the usage level of another computer device 112. A system and method for creating an overall utilization metric for a computer device 112 is described in co-pending and commonly assigned U.S. patent application Ser. No. 12/123,885 entitled, â€œSystem and Method for Determining Overall Utilization.â€ One of ordinary skill in the art will appreciate that the present disclosure contemplates the use of any suitable utilization metric 114 or combination of utilization metrics 114 to describe the current activity level of a computer device 112.

To determine the best computer device 112 available to perform job 104, brokering engine 106, upon receiving a request for a best computer device 112 from client machine 102 to perform job 104 (e.g., as part of a request to place job 104), may receive one or more job criteria 105 associated with job 104 that may define a set of operational characteristics needed for a computer device 112 to be suitable for executing job 104. Brokering engine 106 may further receive one or more utilization criteria 107 associated with job 104 that may define an amount of resource availability (e.g., a level of utilization) needed for a computer device 112 to be suitable for executing job 104. Using job criteria 105 to weed out computer devices 112 that do not have suitable operational characteristics, brokering engine 106 may compile a list of suitable computer devices 112 within active pool 110 to perform job 104. In particular embodiments, the list of suitable computer devices 112 may include only those computer devices 112 within active pool 110 having operational characteristics that satisfy job criteria 105.

Once brokering engine 106 has compiled the list of suitable computer devices 112, brokering engine may compare usage statistics 114 (e.g., an overall utilization value derived from usage statistics 114) for each computer device 112 on the list with utilization criteria 107 to identify which computer device(s) 112 on the list have a suitable level of utilization for performing job 104. As brokering engine 106 compares the utilization level of each computer device 112 on the list against utilization criteria 107, brokering engine 106 may remove from the list computer devices 112 having a utilization level that does not satisfy utilization criteria 107. Once brokering engine 106 has identified the computer device(s) 112 on the list having a suitable level of utilization for performing job 104, brokering engine 106 may provide the list of suitable computer devices 112 to client machine 102 as the list may be considered to include the best computer device(s) 112 for performing job 104. Client machine 102 may then select a computer device 112 from the list and forward job 104 to that computer device 112 for execution.

In particular embodiments, the list of computer device(s) 112 provided to client machine 102 may only include a single computer device 112, such as for example, if utilization criteria 107 specifies that brokering engine 106 should identify the least utilized computer device 112 on the list of suitable computer devices 112. In particular embodiments, the list of computer device(s) 112 provided to client machine 102 may include multiple computer device 112 such as for example, if utilization criteria 107 specifies that brokering engine 106 should identify all computer devices 112 operating within a certain level of utilization. If the list of suitable computer devices 112 returned to client machine 102 contains multiple computer devices 112, client machine 102 may select one of the computer devices 112 from the list using any suitable means of selection. For example, the list of suitable computer devices 112 may be presented to a user who may select a computer device 112 to execute job 104, or client machine 102 may automatically select one of the computer devices 112 on the list to execute job 104.

A job criteria 105 may specify any desired operational characteristic of a computer device 112 needed for computer device 112 to satisfactorily perform job 104. As an example and not by way of limitation, job criteria 105 may include, a desired type or version of software such as an operating system, a desired type of processor, a desired processor speed, a desired amount of RAM, a desired amount of hard drive space, an number of computer devices 112 needed to perform the job, a total amount of physical memory, a total amount of SWAP, an amount of available physical memory, and amount of available SWAP, a minimum amount of time over which computer device 112 must keep a certain amount and type of system resources (e.g., CPU utilization) available to perform job 104, or any other desired operating characteristic. In particular embodiments, one or more job criteria 105 may be predicated on required operational characteristics without which a computer device 112 cannot perform job 104. In particular embodiments, one or more job criteria 105 may be predicated on preferred operational characteristics that are not necessary to perform job 104, but which are preferred for one reason or another. In any case, brokering engine 106 may use job criteria 105 to disqualify unsuitable computer devices 112 from performing job 104, thereby eliminating the need, for example, for brokering engine 106 to determine the utilization level of the unsuitable computer devices 112.

Utilization criteria 107 may be any criteria or combination of criteria that specify a desired level of utilization needed for a computer device 112 to satisfactorily perform job 104. As an example and not by way of limitation, utilization criteria 107 may specify that the least utilized computer device 112 on the list of suitable computer devices should be selected to perform job 104. As an additional example and not by way of limitation, utilization criteria 107 may specify a maximum acceptable level of utilization that represents a threshold level of resource usage, above which brokering engine 106 will determine that a computer device 112 is too busy to perform job 104. However, if the resource usage of a computer device 112 falls below the maximum acceptable utilization level, brokering engine 106 may determine that that computer device 112 has a suitable level of utilization for performing job 104. As an example and not by way of limitation, if the level of utilization for each computer device 112 is expressed as an overall utilization value, the utilization criteria 107 may be a single numeric value against which brokering engine 106 may compare the overall utilization value for a particular computer device 112 to determine whether that computer device 112 is too busy to perform job 104. Depending upon the design of brokering engine 106, two of more utilization criteria 107 may be flexibly combined with one another. For example a first utilization criteria 107 may specify that the least utilized computer device 112 should be selected to perform job 104 while a second utilization criteria identifies a maximum acceptable level of utilization above which brokering engine 106 will determine that a resource 112 is unsuitable to perform job 104 even if that resource 112 is the least active resource 112 available. In any case, brokering engine 106 may use utilization criteria 107 to identify one or more computer devices 112 having a suitable level of utilization for performing job 104.

Depending upon the design of system 100, the job criteria 105 and utilization criteria 107 associated with a job 104 may be specified by a user, derived by brokering engine 106 from the characteristics of job 104, or both. As an example and not by way of limitation, when a user submits a job 104 using client machine 102, the user may specify that the job 104 requires a certain type of processor and a certain amount of free memory and that the user would like the least utilized computer device 112 available to perform the job. As an additional example and not by way of limitation, when brokering engine 106 receives the request to determine the best computer device 112 to perform job 104, brokering engine 106 may determine that job 104 requires a particular program for execution. Accordingly, brokering engine 106 may combine the job criteria 105 that it received from the user with the job criteria 105 that it derived from the characteristics of job 104 to create a complete list of job criteria 105 for narrowing the field of available computer devices 112 in active pool 110 to include only the computer devices 112 whose operational characteristics satisfy job criteria 105. In particular embodiments job criteria 105 may be expressed, for example, in Job Description Language (â€œJDLâ€ ).

Once brokering engine 106 has compiled a list of suitable computer devices 112 whose operational characteristics satisfy job criteria 105, brokering engine 106 may compare the utilization statistics 114 for each computer device 112 on the list against utilization criteria 107 to identify the computer device(s) 112 on the list having a suitable level of activity for performing job 104. Depending upon the design of system 100, the utilization metrics 114 chosen to gauge the activity level of each computer device 112 may be specified by a user, specified by brokering engine 106, or both. As an example and not by way of limitation, when a user submits a job 104 to brokering engine 106, the user may specify that memory usage should be used as the relevant utilization metric 114 to gauge the utilization level of each computer device 112. Depending upon the design of system 100, the user may further specify a mathematical formula for determining a customized aggregated value from a plurality of user-specified utilization metrics 114. In such a case, brokering engine 106 may compare the customized aggregated value for each computer device 112 against utilization criteria 107 to identify the computer device(s) 112 having a suitable level of utilization for performing job 104. Typically, the utilization metrics 114 chosen to gauge the activity level of each computer device 112 should be common across all computer devices 112 to ensure a proper basis of comparison.

In particular embodiments, brokering engine 106 may use one or more default utilizations metrics 114 (e.g., CPU usage, Memory usage, or an overall utilization metric) to gauge the activity level of each computer device 112. Furthermore, brokering engine 106 may supply a default formula for determining an aggregated value from the default utilization metrics 114. As an example and not by way of limitation, brokering engine 106 may identify the least utilized computer device 112 as the computer device 112 with the least percentage of CPU use. If computer device 112 is a multiprocessor computer device 112, the percentage of CPU usage may be derived using the following formula: (total percentage of CPU use)=(Sum of Percentage of CPU use for each CPU)/(number of CPUs). As an additional example and not by way of limitation, brokering engine 106 may determine a percentage of memory (e.g., SWAP memory or physical memory) for a resource using the following formula: (Percentage of Used Memory)=(Memory Used)/(Total Memory Available). One of ordinary skill in the art will appreciate that the above-described formulas and utilization metrics 114 were presented for the sake of explanatory simplicity and will further appreciate that the present disclosure contemplates the use of any suitable number and type of utilization metrics 114 in any suitable formula to determine a numeric measure of utilization for a computer device 112.

Depending upon the design of system 100, brokering engine 106 may receive utilization metrics 114 from all of the computer devices 112 in active pool 110 at regular intervals (e.g., on a regular polling cycle) so that brokering engine 106 has such information immediately available at the time a job 104 is submitted. Alternatively, brokering engine 106 may gather utilization metrics 114 from computer devices 112 after job 104 has been submitted. Brokering agent 106 may receive utilization metrics 114, for example, from a collection agent residing on each computer device 112, though one of skill in the art will appreciate that any suitable means of collection may be used. In any case, brokering engine 106 may use utilization metrics 114 to identify the computer device(s) 112 having a suitable level of utilization for performing job 104 at the approximate moment in time that job 104 is submitted.

In certain situations, it may be the case that no computer device 112 is available in active pool 110 to perform job 104, for example, because none of the computer devices 112 in active pool 110 satisfy job criteria 105, or because all of the computer devices 112 that do satisfy job criteria 105 are too busy to perform job 104 (e.g., if the utilization level for every computer device 112 exceeds the maximum acceptable level of utilization specified by a user as part of utilization criteria 107). In such a case, brokering engine 106 may, for example, postpone performance of job 104 until an appropriate computer device 112 becomes available, or brokering engine 106 may request to provision new computer devices 112 from inactive pool 118 to perform job 104.

When brokering engine 106 provisions a new computer device 112 from inactive pool 118, brokering engine may take whatever steps are necessary to make the new computer device 112 ready to execute job 104. For example, brokering engine may request to turn the new resource on, supply necessary software to the new computer device 112, or perform any other suitable steps needed to enable the new computer device 112 to perform job 104. Once the new resource has been set up to perform job 104, brokering engine 106 may forward job 104 to the new computer device 112 for execution.

In particular embodiments, brokering engine 106 may monitor a group of computer devices 112 as a single entity referred to as a service 128. As an example and not by way of limitation, active pool 110 may include multiple services 128, each including a separate group of computer devices 112. Like computer devices 112, services 128 may be monitored by brokering engine 106 using utilization metrics 114. For example, brokering engine 106 may aggregate the utilization metrics 114 for each computer device 112 included in a particular service 128 into a single value that represents the overall usage of the particular service 128. Brokering engine 106 may then use the aggregated value for each service 128 to determine the least active service 128 in active pool 110.

In particular embodiments, job criteria 105 and utilization criteria 107 may be tailored to services 128 and brokering engine 106 may use the process described above to determine the best service 128 available in active pool 110 to perform a job 104 submitted by client machine 102 or may determine that a new service 128 should be provisioned if no suitable service 128 is available.

FIG. 2 illustrates an example method for dynamic resource provisioning for job placement according to an example embodiment of the present disclosure. The method begins at step 200 where brokering agent 106 receives a request to find a best computer device 112 on which to run a job 104 from client machine 102 including job criteria 105 and utilization criteria 107. The method continues at step 204 where brokering engine 106 determines a list of suitable computer devices 112 for performing job 104 by comparing the operational characteristics of each of the computer devices 112 in pool 110 against job criteria 105 and disqualifying any computer devices whose operational characteristics do not satisfy job criteria 105. The method continues at step 206 where brokering engine 106 compares the utilization metrics 114 gathered from each of the computer devices 112 on the list of suitable computer devices 112 against utilization criteria 107 to identify computer device(s) 112 having a suitable level of utilization for performing job 104. At step 208, the list of computer devices 112 having suitable operational characteristics and a suitable level of utilization for performing job 104 is returned to client machine 102 by brokering engine 106. The method continues at step 210 where client machine 102 forwards job 104 to one of the computer devices 112 on the list of suitable computer devices 112 for execution, after which the method ends.

As mentioned above with respect to FIG. 1, example utilization metrics 114 may include metrics such as those listed in TABLE 1 below.

TABLE 1PlatformMetricDescriptionWindowsPhysical disk:The rate which data is read from Disk Read Bytes the physical disk.per secondWindowsPhysical disk:The rate which data is written to Disk Write Bytes the physical disk.per secondWindowsPhysical disk:The percentage of elapsed time Percentage Disk Timethat the disk drive is busy servicing read or write requests.WindowsPercentage Processor The percentage of elapsed time Timethat the processor spends running a non-idle thread.WindowsMemory:The amount of physical memory Available MBin megabytes available to processes running on the computer.WindowsMemory:The rate at which pages are read Pages per Secondfrom or written to the disk to resolve hard page faults.WindowsMemory:The percentage of total possible Percentage physical memory in use for Committed Bytes in which space has been reserved in Usethe paging file if it needs to be written to the disk.WindowsNetwork Interface:The rate at which bytes are sent Bytes Total per and received over each network Secondadapter including message framing characters.WindowsNetwork Interface:The estimated current bandwidth Current Bandwidthof the network interface in bits per second. For example, 10000 represents 10 Mbps.WindowsNetwork Interface:The rate at which packets are sent Packets per Secondand received on the network interface.Windows Uptime PercentageThe percentage of time a cluster Microsoft has been evaluated as being in an Cluster Server â€œUpâ€ state.â€œMSCSâ€ ClusterWindows Uptime PercentageThe percentage of time a cluster MSCS Nodenode has been evaluated as being in an in an â€œUpâ€ state.Windows Uptime PercentageThe percentage of time a cluster MSCS Re-resource group has been source Groupevaluated as being in an in an â€œUpâ€ state.Windows Uptime PercentageThe percentage of time a cluster MSCS resource has been evaluated as Resourcebeing in an in an â€œUpâ€ state.Windows Uptime PercentageThe percentage of time a cluster MSCS Network network interface has been Interfaceevaluated as being in an in an â€œUpâ€ state.Windows Uptime PercentageThe percentage of time a cluster MSCS Networknetwork has been evaluated as being in an in an â€œUpâ€ state.Windows Uptime PercentageThe percentage of time a cluster Veritas Clusterhas been evaluated as being in an in an â€œUpâ€ state.Windows Uptime PercentageThe percentage of time a cluster Veritas Cluster node has been evaluated as beingNodein an in an â€œUpâ€ state.Windows Uptime PercentageThe percentage of time a cluster Veritas Cluster service group has been evaluated Service Groupas being in an in an â€œUpâ€ state.Windows Uptime PercentageThe percentage of time a cluster Veritas Cluster resource has been evaluated as Resourcebeing in an in an â€œUpâ€ state.LinuxCPU Group:The percentage of elapsed time Total Usage that the process spends to run Percentage Useruser instructions.LinuxCPU Group:The percentage of elapsed time Total Usage that the processor spends to run Percentage Systemoperating system instructions.LinuxCPU Group:The fifteen minute load average.Load Average FifteenLinuxDisk Group:The rate which data is read from Bytes per Secondand written to the physical disk.LinuxDisk Group:The rate which data is read from Reads per Secondthe physical disk.LinuxDisk Group:The rate which data is written to Writes per Secondthe physical disk.LinuxNetwork Group:The rate at which the network TCP Operations interface is receiving packets.Packets Received per SecondLinuxNetwork Group:The rate at which the network TCP Operations interface is sending packets.Packets Sent per SecondLinuxFile System Group:Available storage space in the Free Spacefile system in kilobytes.LinuxFile System Group:Used storage space in the file Used Spacesystem in kilobytes.Sun Solaris Uptime PercentageThe percentage of time a cluster Veritas Clusterhas been evaluated as being in an in an â€œUpâ€ state.Sun Solaris Uptime PercentageThe percentage of time a cluster Veritas Cluster node has been evaluated as being Nodein an in an â€œUpâ€ state.Sun Solaris Uptime PercentageThe percentage of time a cluster Veritas Cluster service group has been evaluated Service Groupas being in an in an â€œUpâ€ state.Sun Solaris Uptime PercentageThe percentage of time a cluster Veritas Cluster resource has been evaluated as Resourcebeing in an in an â€œUpâ€ state.SolarisCPU Group:The percentage of elapsed time Total Usage that the process spends to run Percentage Useruser instructions.SolarisCPU Group:the percentage of elapsed time Total Usage that the processor spends to run Percentage Systemoperating system instructions.SolarisCPU Group:The fifteen minute load average.Load Average FifteenSolarisDisk Group:The rate which data is read from Bytes per Secondand written to the physical disk.SolarisDisk Group:The rate which data is read from Reads per Secondthe physical disk.SolarisDisk Group:The rate which data is written to Writes per Secondthe physical disk.SolarisNetwork Group:The rate at which the network TCP Operations interface is receiving packets.Packets Received per SecondSolarisNetwork Group:The rate at which the network TCP Operations interface is sending packets.Packets Sent per SecondSolarisFile System Group:Available storage space in the Free Spacefile system in kilobytes.SolarisFile System Group:Used storage space in the file Used Spacesystem in kilobytes.VMware ESX VMware ESX 3.0 The number of kilobytes read ServerServer Disk:from the disk per second.KB Read per SecondVMware ESX VMware ESX 3.0 The number of kilobytes written ServerServer Disk: KB to the disk per second.Written per SecondVMware ESX VMware ESX 3.0 VMThe percentage of elapsed time ServerCPU: CPU Usage that the processor uses to run the (percentage)virtual machine.VMware ESX VMware ESX 3.0 VMThe percentage of the total ServerMemory:amount of physical memory used Memory Usageby the virtual machine.(percentage)VMware ESX VMware ESX 3.0 VMThe amount of memory available ServerMemory: Available for use by the virtual machine in Memory (KB)kilobytes.VMware ESX VMware ESX 3.0 VMThe number of kilobytes read ServerDisk:from the virtual machine disk per KB Read per Secondsecond.VMware ESX VMware ESX 3.0 VMThe number of kilobytes written ServerDisk: KB to the virtual machine disk per Written per Secondsecond.VMware ESX VMware ESX 3.0 VMThe number of kilobytes trans-ServerNetwork: mitted (total amount of data sent KB Transmitted per and received) by the virtual Secondmachine over the network per second.VMware ESX VMware ESX 3.0 VMThe number of kilobytes received ServerNetwork: KB by the virtual machine over the Received per Secondnetwork per second.IBM AIXMemory Group: The percentage of physical Percentage Free memory free on the system.Physical MemoryIBM AIXCPU:The total percentage of CPU Percentage Totalutilization on the system.IBM AIXCPU Group:Total CPU utilization due to user Total Usage activity on the system.Percentage UserIBM AIXCPU Group:Total CPU utilization due to OS Total Usage activity on the system.Percentage SystemIBM AIXCPU Group: Load The fifteen minute load-average Averages Fifteenof the system.IBM AIXDisk Group:The number of bytes read and Bytes per Secondwritten per second by the disk.IBM AIXDisk Group: Reads The average number of reads per per Second (average)second by the disk.IBM AIXDisk Group: The average number of writes per Writes per Secondsecond by the disk.(average)IBM AIXNetwork Group: TCP The number of packets received Operations Packets per second by the network Received per Secondadapter.IBM AIXNetwork Group: TCP The number of packets sent per Operations Packets second by the network adapter.Sent per SecondIBM AIXFile System Group:Amount of free space (in bytes) Free Spaceof the file system.IBM AIXFile System Group:Amount of used space (in bytes) Used Spaceof the file system.IBM AIXCPU Group: Total The percentage of CPU used by Usage Percentage Idlethe system idle process.HP UNIXMemory Group: The percentage of physical Percentage Free memory free on the system.Physical MemoryHP UNIXCPU:The total percentage of CPU Percentage Totalutilization on the system.HP UNIXCPU Group: Total Total CPU utilization due to user Usage Percentageactivity on the system.UserHP UNIXCPU Group: Total Total CPU utilization due to OS Usage Percentageactivity on the system.SystemHP UNIXCPU Group: Load The fifteen minute load-average Averages Fifteenof the system.HP UNIXCA Disk Group:The number of bytes read and Bytes per Secondwritten per second by the disk.HP UNIXDisk Group: Reads The average number of reads per per Second (average)second by the disk.HP UNIXDisk Group:The average number of writes per Writes per Secondsecond by the disk.(average)HP UNIXNetwork Group: TCP The number of packets received Operations Packetsper second by the network Received per Secondadapter.HP UNIXNetwork Group: TCP The number of packets sent per Operations Packetssecond by the network adapter.Sent per SecondHP UNIXFile System Group:Amount of free space (in bytes) Free Spaceof the file system.HP UNIXFile System Group:Amount of used space (in bytes) Used Spaceof the file system.HP UNIXCPU Group: Total The percentage of CPU used by Usage Percentage Idlethe system idle process.WindowsAgent Stats:Total number of metrics of an Metric CountIntroscope agent collectsWindowsBackends:Average milliseconds taken for a Average Response backend component to respond to Time (ms)a requestWindowsBackends:Number of backend errors per Errors Per Intervalpoll intervalWindowsBackends:Number of times the backend Stall Countinvocation has been stalledWindowsFrontends:Average number of milliseconds Average Response taken for a frontend component Time (ms)to respond to a requestWindowsFrontends:Number of frontend errors per Errors Per Intervalpoll intervalWindowsFrontends:Number of times the frontend Stall Countinvocation has been stalledWindowsFrontends:Number of open connections to a Connection Countfrontend componentWindowsCPU: Utilization % Aggregate % CPU utilization(aggregate)WindowsCPU: Utilization % Application's % CPU utilization(process)WindowsGC Heap:Total JVM heap size in bytesBytes TotalWindowsGC Heap:Bytes in use by JVMBytes In UseWindowsEJB:Average number of milliseconds Average Response taken for an EJB component to Time (ms)respond to a requestWindowsEJB:Number of EJB errors per poll Errors Per IntervalintervalWindowsEJB:Number of times the EJB Stall Countinvocation has been stalledWindowsJSP:Average number of milliseconds Average Response taken for an JSP component to Time (ms)respond to a requestWindowsJSP: Responses Per Number of responses per poll Interval interval for a JSP componentWindowsJSP:Number of JSP errors per poll Errors Per IntervalintervalWindowsJSP:Number of times the JSP Stall Countinvocation has been stalledWindowsServlets:Average number of milliseconds Average Response taken for an Servlet to respond to Time (ms)a requestWindowsServlets: Responses Number of responses per poll Per Intervalinterval for a Servlet componentWindowsServlets:Number of Servlet errors per poll Errors Per IntervalintervalWindowsServlets:Number of times the Servlet Stall Countinvocation has been stalledWindowsSockets:Number of sockets opened per Opens Per Intervalpoll intervalWindowsSockets:Number of sockets closed per Closes Per Intervalpoll intervalWindowsSockets:Number of socket accepts per Accepts Per Intervalpoll interval

Although the present disclosure has been described in several embodiments, a myriad of changes, substitutions, and modifications may be suggested to one skilled in the art, and it is intended that the present disclosure encompass such changes, substitutions, and modifications as fall within the scope of the present appended claims.