The present invention relates to the field of computing. More particularly, the present invention relates to the field of computing where a client requests activity by a server.
In a network environment, an initiating computer (i.e., a client) sends a message to another computer (i.e., a server) requesting that the server perform some action. The World Wide Web employs this technique. A web browser (i.e., the client) requests web pages from a web server (i.e., the server).
This technique has been widely used because it is easy to model and because it clearly distributes responsibilities between the client and the server. The server can perform only operations which the server exposes to the network. Upon an exception or undesirable results, the server contacts the client for instructions. Often, it would be desirable to resolve the exception or undesirable results at the server without further contact with the client.
The present invention comprises a method of processing a request by a server computer system. In an embodiment, the method begins with a step of receiving an object. The object comprises the request, an exception criterion, and a recovery operation. The method continues with a step of determining that the server computer system is unable to fulfill the request due to an exception which meets the exception criterion. The method concludes with a step of applying the recovery operation to resolve the exception.
In an embodiment, the server computer system comprises a resource management system for a shared computing environment and the object comprises a resource request object. The resource request object comprises a resource request, the exception criteria, and the recovery operation. The resource request indicates resources requested from the shared computing environment by a start time and for a time period.
These and other aspects of the present invention are described in more detail herein.
The present invention is described with respect to particular exemplary embodiments thereof and reference is accordingly made to the drawings in which:
According to an embodiment, the present invention comprises a method of managing resources in a shared computing environment (e.g., a utility data center). According to another embodiment, the present invention comprises a system for managing resources in a shared computing environment. According to yet another embodiment, the present invention comprises a method of processing a request by a server computer system.
An embodiment of a shared computing environment which employs a method of managing resources of the present invention is illustrated schematically in
A resource pool of the shared computing environment 100 comprises resource types. According to an embodiment, the resource types comprise the processors 114, memory, and storage. According to another embodiment, the resource types comprise the servers 102 and the disk arrays 104. Depending upon the resource type, the client requests a quantity or a capacity of a resource type. For example, the client may request two processors, 500 MB of memory capacity, and 100 GB of storage capacity. Alternatively, the client may request the resources in some other way such as one or more servers and one or more disk arrays.
An embodiment of the method of managing resources in the shared computing environment 100 of the present invention is illustrated as a flow chart in
In an embodiment, the method 200 continues with a second step 204 of allocating the resources to the client (i.e., a resource allocation). The second step 204 does not assign instances of the resources. Rather, the second step 204 reduces an available resource pool by the resources for the time period which the client requested the resources. The second step 204 amends the resource request object so that it notes the resource allocation. In another embodiment, the method 200 does not perform the second step 204.
The method 200 continues with a third step 206 of assigning instances of the resources to the client. In a fourth step 208, the method 200 deploys the instances of the resources for the client by a start time. In a fifth step 208, the method sends a resource access handle to the client. The resource access handle provides the client with access to the resources by the start time. In a sixth step 210, the method tracks the assigned resources and other state information by amendments to the resource request object.
By tracking the resource request through amendment of the resource request object, the resource request object contains a context for the resource request. Thus, information regarding a particular resource request can be obtained by requesting the information from the resource request object or accessing the resource request object for the information. This information may be used by the resource provider or the client. For example, the resource provider may use this information in a migration between servers or storage resources. Or, for example, the client may use the information to determine whether the resources are performing adequately. As the client uses the resources for the time period, the resource request object is occasionally amended to indicate a current state. The current state updates the information held by the resource request object and may include a checkpoint for restart upon failure or migration. When the time period expires, the resources are returned to the resource pool.
An embodiment of a resource management system of the present invention is illustrated schematically in
In operation, the advertisement subsystem 318 (
In an embodiment, the client 302 submits a resource query 322 to the resource information service 304 in order to identify a resource provider with the best offering. The resource query 322 identifies desired capabilities such as resource types, resource quantities or capacities, available software packages, available operating systems, or available management platforms. The resource information service 304 returns a resource provider's contact information 324 which best meets the client's resource request. For example, the resource information service may return the resource management system 300 as the best resource provider for the client's request.
The client 302 prepares a complete resource request description and uses it to prepare a resource request object. The resource request object includes requested resource types, requested quantities or capacities of the resource types, a resource topology, and the time period for the request. The resource topology comprises network bandwidth (e.g., a LAN bandwidth, a SAN bandwidth, and a bandwidth for an Internet connection). The resource topology may also include a more detailed connection relationship between the resources. According to embodiments, the resource request object also includes optional information such as resource dependency descriptions, workload profile, alternative resource types, alternative resource quantities or capacities, alternative time periods, SLA (service level agreement) requirements, or policy information (e.g., recovery operations for responding to exceptions).
In an embodiment, the client 302 sends a message 326 containing the resource request object to the reservation subsystem 306. In another embodiment, the client 302 sends a reference to the reservation subsystem 306, which then accesses the resource request object. In this embodiment, the client 302 may retain the resource request object. Alternatively, the client 302 may send the resource request object to the resource management system 300 which places the resource request object in an accessible location for access by subsystems of the resource management system 300. The reservation subsystem 306 determines that the shared computing environment can fulfill the resource request contained in the resource request object based upon a comparison of the resource request to available resources (i.e., non-reserved resources). The reservation subsystem 306 passes the resource request object or a reference to the resource request object to the allocations subsystem 308, which allocates resources from the resource pool 318 for the time period identified in the resource request object. The resource allocation does not assign instances of the resources. Rather, the resource allocation reduces available resources by the resource quantities and capacities for the time period identified in the resource request object. The resource allocation subsystem 308 amends the resource request object to identify the resource allocation.
Just prior to when the resource request becomes due, the allocation subsystem 308 forwards the resource request object or a reference to the resource request object to the assignment subsystem 310. The particular time chosen for forwarding the resource request object or the reference to the resource request object is selected on the basis of providing sufficient time for accomplishing assignment of the resources. The assignment subsystem 310 assigns instances of the resources to the resource request object, amends the resource request object with an identification of the instances of the resources, and forwards the resource request object or a reference to the resource request object to the deployment subsystem 312.
The deployment subsystem 312 deploys the resources to meet the requirements of the resource request object. This may include configuring a topology for the resources or initializing the resources. The deployment subsystem 312 sends a message 328 containing a resource access handle to the client 302. The resource access handle provides access to the resources by the start time of the time period in the resource request object. The deployment subsystem 312 amends the resource request object to identify the deployment and forwards the resource request object or a reference to the resource request object to the operations control subsystem 314.
As the client 302 employs the resources, the operations control subsystem 314 monitors and controls the resources. This may include implementing policy contained in the resource request object. Occasionally, the operations control subsystem 314 amends the resource request object to identify a state of the resources.
An embodiment of a workflow for a resource request of the present invention is illustrated as a state diagram in
If the reservation state 402 accepts the reservation, the reservation state 402 can also allocate resources to the reservation and amend the resource request object to identify the resource allocation. Alternatively, the resource request or a reference to the resource request object can be passed to an allocation state 404, which allocates the resources for the reservation and amends the resource request object to identify the resource allocation. The resource request waits in the reservation state 402 or the allocation state 404 until just before the resource become due.
Before the resources become due, the resource request passes to an assignment state 406. The assignment state 406 assigns instances of the resources to the resource request and amends the resource request object to identify the instances of the resources. The instances of the resources are selected from a pool of available resources. The resource request then passes to a deployment state 408.
The deployment state 408 creates a specific arrangement of the assigned resources to meet the client's requirements and amends the resource request object to identify the specific arrangement. The resource request then passes to an activation state 410.
The activation state 410 gives the client access to the resources by sending the resource access handle to the client. The activation state 410 amends the resource request object to indicate that the resources have been activated for the client. The resource request then passes to a use state 412 in which the client makes use of the resources. While the resources are in the use state 412, they are unavailable for allocation. In the use state 412, the resource request object is occasionally amended to identify a current state of the resources.
Eventually, the time period identified in the resource request object expires and the resource request object passes to a release state 414. The release state 414 moves the resource back into an available resource pool so that the resources can be allocated and assigned to other future requests. The release state 414 can also include cleaning states from the resources after use for security purposes. The resource request then proceeds to a termination state 416 which ends the client's association with the resource management system for this resource request.
An embodiment of a workflow for a resource request of the present invention is illustrated as a sequence diagram in
The workflow 500 continues with the client 502 creating the resource request object 504 in a first action 512. The resource request object 504 identifies the client's requirements and may include workflow instructions. According to an embodiment, the workflow instructions implement pre-defined policies (i.e., recovery operations) upon encountering workflow exceptions.
The client 502 sends a message 513 to the reservation subsystem 506 which references the resource request object 504. In an embodiment, the client sends the resource request object 504 to the resource management system 505. In another embodiment, the client 502 retains the resource request object. The reservation subsystem 506 reads the resource request object in a read operation 514 to determine the resource request contained in the resource request object. In an embodiment, the reservation subsystem 506 allocates resources to the resource request object and indicates the resource allocation by an amendment 516 to the resource request object 504. In another embodiment, an allocation subsystem (not shown) allocates the resources and amends the resource request object 504. In an embodiment, the reservation subsystem 506 sends an acknowledgment 518 indicating that a reservation has been accepted.
In an alternative embodiment, the reservation subsystem 506 rejects the reservation. In such an embodiment, the reservation subsystem 506 may forward a reference to the resource request object 504 to another resource management service in accordance with the workflow instructions. Alternatively, the reservation subsystem 506 may notify the client 502 of the rejection.
Preferably, if the reservation subsystem accepts the reservation, the reservation subsystem waits until a time 520 which is just prior to when the reservation becomes due. The reservation subsystem 506 then forwards a reference to the resource request object 504 to the assignment subsystem 508 in a message 522. Alternatively, the reservation subsystem sends the message 522 including the reference to the resource request object at an earlier time.
In an embodiment, the assignment subsystem 508 reads the resource request object in a read operation 524, assigns the resources according to the resource request object, and notes the assignment by amending the resource request object 504 in a message 526. In an embodiment, the assignment subsystem 508 notifies the reservation subsystem 506 of the assignment in a message (not shown).
In an alternative embodiment, the assignment subsystem 508 fails to assign the resources 508. For example, while the reservation subsystem 506 may have reserved sufficient resources for the resource request, by the time of allocation there might be insufficient resources available for the resource request due to an unexpected failure of some resources. According to such an embodiment, the assignment subsystem 508 may try to assign the resource according to the workflow instructions again. Alternatively, the assignment subsystem 508 may notify the reservation subsystem 506 of the failure to assign resources. In response, the reservation subsystem may forward a reference to the resource request object to an alternative resource management service that can assign the resources according to the workflow instructions in a message (not shown).
If the assignment subsystem 508 assigns the resources to the resource request object 504, the assignment subsystem 508 forwards a reference to the resource request object 504 to the deployment subsystem 510 in a message 527. The deployment subsystem 510 reads the resource request from the resource request object in a read operation 528, activates the resource for the client, and configures a topology for the client. The deployment subsystem 510 provides a resource access handle to the client in a message 530.
The resource access handle provides the client 502 with the ability to use the resources by the start time for the time period indicated in the resource request object 504. The client 502 makes use of the resources during the time period. Occasionally during the use, an operations control subsystem (not shown) updates the resource request object with a current state of the resources. At the end of the time period, the client releases the resources via a message 534 to the deployment system 510. The deployment subsystem 510 returns the resources to an available resource pool and notifies the assignment subsystem 510 of the release via a message 538. The client 502 terminates the session via a message 540 to the resource request object 504, which deletes the resource request object and the session state information held by the resource request object.
An alternative embodiment of a workflow is illustrated in
The workflow 600 continues with the client 602 submitting the resource request object to the reservation subsystem 606 in a transmission 612. The reservation subsystem 606 determines whether to accept the resource request. If the resource request is accepted, the reservation subsystem 606 notifies the client 602 of the acceptance in a transmission 614 and amends the resource request object to indicate the acceptance. If the resource request is rejected, the reservation subsystem 606 forward the resource request object to an alternative resource management system (not shown) according to the workflow instruction contained in the resource request object. If the workflow instructions do not indicate an alternative resource management system and the resource request is rejected, the reservation subsystem 606 returns the resource request object to the client 602.
Upon accepting the resource request, the reservation subsystem 606 preferably waits until a time 615 just prior to the reservation becoming due and then forwards the resource request object to the assignment subsystem 608 in a transmission 618. Alternatively, the reservation subsystem forwards the resource request object to the assignment subsystem 608 at an earlier time.
Preferably, the assignment subsystem 608 assigns the resources according to the resource request contained in the resource request object, amends the resource request object to indicate the assignment, and forwards the resource request object to the deployment subsystem 610 in a transmission 620. Alternatively, the assignment subsystem 608 fails to assign the resource, notes the failure in the resource request object, and responds to the failure according to the workflow instructions or returns the resource request object to the client 602.
The deployment subsystem 610 activates the resources 622 for the client 602 in a transmission 624. The activation of the resources 622 includes configuring a topology for the resources 622. The deployment subsystem 610 also provides a resource access handle to the client 602 in a transmission 626. The resource access handle allows the client 602 to use the resources 622 by the start time and for the time period indicated in the resource request object. The client 602 makes use of the resources 622 during the time period indicated a series of transmissions 628. At the end of the time period, the client 602 releases the resources 622 in a transmission 630.
The resource management system 300 (
A second benefit is that the resource management system automatically coordinates multiple resource requests from different clients. Each resource request moves through its workflow proceeding to an appropriate subsystem according to a state of the workflow, which distributes a processing load over the subsystems.
A third benefit is that, because each resource request object accessed by a resource management system may include workflow instructions, each client can provide independent instructions for its workflow to the resource management system.
A fourth benefit is that resource pools may be coupled to an enterprise wide grid or a public resource grid in which clients request resource from various resource management services. This allows resource providers to join, leave, or change their offerings dynamically. It also allows clients to evaluate offerings and to choose a preferred resource provider.
A fifth benefit is that a workflow for a resource request centers on the resource request object which holds a context of the resource request including a current state of the resource request. This allows the resource request to be modified by the client by contacting the resource request object, possibly in response to a workflow exception or possibly due to a changed need for the resources. It also allows checkpointing of a resource request in execution and transferring the resource request to another resource management system by forwarding the resource request object or a reference to the resource request object.
A sixth benefit is that the present invention modularizes various tasks associated with a resource request. This includes having an external resource information service which provides clients with a central location for evaluating various resource providers. This also includes having various subsystems in the resource management system to handle different tasks associated with a resource request.
As discussed above, embodiments of the resource request object include workflow instructions for handling workflow exceptions. These embodiments employ a method of processing a request by a server computer system of the present invention which may be used for requesting resources or in other situations such as business processes or e-commerce applications.
An embodiment of a method of processing a request by a server computer system of the present invention is illustrated as a flow chart in
In an embodiment, the method continues with a second step 704 of encountering the exception while processing the data. In this embodiment, the method concludes with a third step 706 of applying the recovery operation contained in the object. In another embodiment, the second step 704 encounters a particular exception within the first through nth exceptions. In this embodiment, the third step 706 applies a particular recovery operation within the first through nth recovery operations corresponding to the particular exception. For example, if the server encounters the second exception, the server applies the second recovery operation to resolve it. In yet another embodiment, the second step 704 encounters one of a group of exceptions. In this embodiment, the third step 706 applies a recovery operation which applies to any of the group of exceptions.
An embodiment of a method of processing a request by a server computer system of the present invention is illustrated as a sequence diagram in
An exemplary embodiment of a method of processing a request by a server computer system involves a web browser (i.e., a client), a first web server, and a second web server. The user of the web browser wants an index from the first web server but, if the index is unavailable from the first web server, the user wants an index from the second web server. The web browser sends a request to the first web server, which may take the form:
GET/index.html HTTP 1.1
User-Agent: Mozilla/4.0
Host: “first web server”
On-Exception-404: GET/index.html (Host: “second web server”)
Connection: Keep-Alive
If the first web server is able to respond correctly with the index, it does so. If the first web server encounters an error (indicated by code 404), the first web server implements recovery operation indicated in the line “On-Exception-404” and forwards the request for an index to the second web server.
Another recovery operation may be nested within a first recovery operation. For example, in the exemplary embodiment above, a second “On-Exception-404” may follow “Host: “second web server,” which identifies a third web server from which to obtain the index upon encountering an error (indicated by code 404) at the second web server.
An embodiment of a method of processing a resource request by a resource management system of the present invention is illustrated as a flow chart in
An embodiment of the resource request object is illustrated as a class diagram in
The method 900 (
An embodiment of first and second resource management systems which employ the method of processing a resource request of the present invention are illustrated schematically in
A first exception may occur when the resource request object is within a reservation state of a resource request workflow for the first resource management system 1102. The first exception is a determination by the first resource management system 1102 that available resources for the requested time period are insufficient to fulfill the resource request. According to an embodiment, the workflow instructions include a recovery operation for this exception which instructs the first resource management system 1102 to attempt to fulfill an alternative resource request. According to another embodiment, the workflow instructions include a recovery operation for this exception which instructs the first resource management system 1102 to forward the resource request object to the second resource management system 1104. According to another embodiment, the workflow instructions include a recovery operation which instructs the first resource management system 1102 to return the resource request object to the client 1108. According to another embodiment, the workflow instructions include a recovery operation which instructs the first resource management system 1102 to propose one or more alternatives to the client.
A second exception may occur when the resource request object is within an assignment state of the resource request workflow. The second exception is a determination by the first resource management system 1102 that there are insufficient resources to fulfill the resource request. According to embodiments, the workflow instructions include recovery operations which provide various alternative workflows for resolution of the lack of sufficient resources to fulfill the resource request.
Various exceptions may occur during use of the resources by the client. According to embodiments, the workflow instructions provide instructions for resolving these exceptions.
According to an embodiment, each subsystem that receives the resource request object or a reference to the resource request object includes a capability to interpret the workflow instructions included in the resource request object. According to an embodiment, a subsystem accomplishes this capability through a process launched for each resource request object that the subsystem receives. The process comprises a workflow interpreter and a messaging capability. When a subsystem determines that the resource request object or a reference to it is to be sent to another subsystem, the process gathers a current state of the resource request and amends the resource request object to include the current state. The process then terminates. When the other subsystem receives the resource request object or the reference to it, the other system launches another process which includes the workflow interpreter and the messaging capability.
According to an embodiment, the workflow is implemented as a finite state machine. The finite state machine comprises a set of states which describe the workflow, a set of transitions described in the workflow, and a current state of the workflow. The current state comprises a minimum amount of information that needs to be passed with the resource request object when it is passed from a first subsystem to a second subsystem.
An exemplary set of workflow instructions is provided in
The foregoing detailed description of the present invention is provided for the purposes of illustration and is not intended to be exhaustive or to limit the invention to the embodiments disclosed. Accordingly, the scope of the present invention is defined by the appended claims.
Number | Name | Date | Kind |
---|---|---|---|
763119 | Sawyer | Jun 1904 | A |
4126895 | Weemaes | Nov 1978 | A |
4495562 | Yamaji | Jan 1985 | A |
5077661 | Jain | Dec 1991 | A |
5109486 | Seymour | Apr 1992 | A |
5287552 | Sasuta | Feb 1994 | A |
5590328 | Seno | Dec 1996 | A |
5603029 | Aman | Feb 1997 | A |
5630133 | Hotea | May 1997 | A |
5634127 | Cloud et al. | May 1997 | A |
5687372 | Hotea | Nov 1997 | A |
5701482 | Harrison | Dec 1997 | A |
5729710 | Magee et al. | Mar 1998 | A |
5745694 | Egawa et al. | Apr 1998 | A |
5758029 | Hall | May 1998 | A |
5889989 | Robertazzi | Mar 1999 | A |
6058426 | Godwin et al. | May 2000 | A |
6112243 | Downs | Aug 2000 | A |
6141759 | Braddy | Oct 2000 | A |
6185609 | Rangarajan et al. | Feb 2001 | B1 |
6237009 | Waldo et al. | May 2001 | B1 |
6256740 | Muller et al. | Jul 2001 | B1 |
6272528 | Cullen et al. | Aug 2001 | B1 |
6289446 | Nilsson | Sep 2001 | B1 |
6347253 | Fujita | Feb 2002 | B1 |
6349343 | Foody et al. | Feb 2002 | B1 |
6370560 | Robertazzi | Apr 2002 | B1 |
6401121 | Yoshida | Jun 2002 | B1 |
6434555 | Frolund et al. | Aug 2002 | B1 |
6438573 | Nilsen | Aug 2002 | B1 |
6442552 | Frolund et al. | Aug 2002 | B1 |
6477558 | Irving | Nov 2002 | B1 |
6510164 | Ramaswamy | Jan 2003 | B1 |
6594774 | Chapman et al. | Jul 2003 | B1 |
6665867 | Ims et al. | Dec 2003 | B1 |
6748554 | Jin et al. | Jun 2004 | B2 |
6760736 | Waldo et al. | Jul 2004 | B2 |
6847974 | Wachtel | Jan 2005 | B2 |
6857085 | Ruhlen et al. | Feb 2005 | B1 |
7003762 | Krueger | Feb 2006 | B2 |
7035959 | Umezu et al. | Apr 2006 | B2 |
7127511 | Tonouchi | Oct 2006 | B2 |
7272640 | Kazemi et al. | Sep 2007 | B1 |
7328406 | Kalinoski et al. | Feb 2008 | B2 |
7523348 | Anand et al. | Apr 2009 | B2 |
7558864 | Kalantar et al. | Jul 2009 | B2 |
7644410 | Graupner et al. | Jan 2010 | B1 |
7647523 | Lindquist et al. | Jan 2010 | B2 |
7673054 | Bozak et al. | Mar 2010 | B2 |
7689971 | Peng et al. | Mar 2010 | B2 |
20010029550 | Endo et al. | Oct 2001 | A1 |
20020013835 | Umezu et al. | Jan 2002 | A1 |
20020073114 | Nicastro et al. | Jun 2002 | A1 |
20020194181 | Wachtel | Dec 2002 | A1 |
20020194251 | Richter et al. | Dec 2002 | A1 |
20030046393 | Mueller | Mar 2003 | A1 |
20030055968 | Hochmuth et al. | Mar 2003 | A1 |
20030208509 | Komine et al. | Nov 2003 | A1 |
20030212792 | Raymond | Nov 2003 | A1 |
20040015851 | Newhook et al. | Jan 2004 | A1 |
20040064558 | Miyake | Apr 2004 | A1 |
20040064570 | Tock | Apr 2004 | A1 |
20040139147 | Duquenne et al. | Jul 2004 | A1 |
20040153841 | Beck | Aug 2004 | A1 |
20040221203 | Ruhlen et al. | Nov 2004 | A1 |
20050015641 | Alur et al. | Jan 2005 | A1 |
20050182843 | Reistad et al. | Aug 2005 | A1 |
20050193225 | Macbeth et al. | Sep 2005 | A1 |
20070250839 | Van Der Sanden et al. | Oct 2007 | A1 |
20080052718 | Hundscheidt | Feb 2008 | A1 |
20080319857 | Dobbins et al. | Dec 2008 | A1 |