CN1954295A - Computer system, method, and program for controlling execution of job in distributed processing environment - Google Patents
Computer system, method, and program for controlling execution of job in distributed processing environment Download PDFInfo
- Publication number
- CN1954295A CN1954295A CNA2005800154954A CN200580015495A CN1954295A CN 1954295 A CN1954295 A CN 1954295A CN A2005800154954 A CNA2005800154954 A CN A2005800154954A CN 200580015495 A CN200580015495 A CN 200580015495A CN 1954295 A CN1954295 A CN 1954295A
- Authority
- CN
- China
- Prior art keywords
- resource
- computer system
- grid
- request
- local
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5072—Grid computing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer And Data Communications (AREA)
- Multi Processors (AREA)
Abstract
The present invention provided a scalable system easily compatible with modification of the system configuration and size enlargement/contraction in a wide-range distributed system unifying and utilizing the grid system groups. A grid server (100) of each grid system constituting the wide-range distributed system includes: a scheduler (110) for allocating a job to resource means including a computer constituting the local system (local resource) and other grid system on the network (network resource); and a resource agent (120) for relaying the communication between the scheduler (110) and the resourcemeans. The resource agent (120) manages the resource means information, receives a job execution request by the scheduler (110) in place of the resource means to which the job has been allocated, and executes the job execution request to the resource means according to the condition of the resource means.
Description
Technical field
The present invention relates to grid (grid) and calculate, particularly synthetically control method and this system architecture of a plurality of computing systems.
Background technology
In recent years, everybody pay close attention to gradually the infosystem of different types of (different mode mixes and exists) that comprehensive utilization connects by network, promptly be called the technology of grid computing.Utilize this technology, computer resource such as the ability of CPU and data-carrier store in a plurality of computing machines on can community network is used as 1 virtual high performance computing machine.Owing in many computing machines, carry out parallel processing, therefore,, can carry out a large amount of processing at high speed although one one performance is very low.
Also have, we consider to carry out virtual to the computer resource that is present in the distributed processing system(DPS) group that is connected on the wide area network, make usefulization of computer resource in the distributed processing system(DPS) group that is present in each other, and make it shared, thereby effectively utilize this resource.Here, so-called distributed processing system(DPS) is meant a plurality of diversified computer resource that is connected on the network is managed as 1 group, carries out the system that load disperses and dispatches (scheduling).So-called distributed processing system(DPS) group means a distribution type disposal system that is present on a plurality of wide area networks (distributed network).In the following description, distributed processing system(DPS) recorded and narrated each computing system in the grid computing environment that is virtualized for resource (below, abbreviate grid system as).
In order to construct the distributed system of this wide area that the computer resource that is present in the grid system group is fully utilized, the computer resource in will managing each grid system, also need the structure of managing gridding set of systems integral body.Therefore, all the time, the load that constantly research, exploitation are used to carry out the computer resource of this grid system group disperses and first scheduler (meta-scheduler) of scheduling.Organizational form with grid system group in the past of first scheduler mainly is divided into totally 3 kinds of concentrated schemes (Centralized Scheme), classification type scheme (HierarchicalScheme), distributed schemes (Distributed Scheme) (for concentrated scheme and classification type scheme, for example see also non-patent literature 1, for distributed schemes, for example see also non-patent literature 2).
Figure 13 illustrates the system architecture based on the concentrated scheme synoptically.
In the concentrated scheme, in first scheduler of all central servers of managing gridding set of systems, concentrated all grid system information, make the decision of scheduling by this yuan scheduler.In site-local (local divider (dispatcher)) as each grid system, the decision that does not work as dispatcher, but carry out the operation of first scheduler input (dropping into the The Ru) from central server, afterwards, operation is finished and the information of resource status (empty processor etc.) is notified to first scheduler.Changed under the situation of system architecture having appended new grid system etc., be reflected in the scheduling of sending, needed the step of utilizing central server to carry out by first scheduler in order to make it.
Figure 14 is the system architecture that generally illustrates based on classification type scheme.
In classification type scheme, between central server and each site-local, share the processing of scheduling.From first scheduler of central server after the scheduler of site-local has been submitted operation to, first scheduler does not need directly related with this operation.If after having submitted operation to, just both other site-local can have the free time, the execution of each operation also is to carry out in having sent each site-local of this operation.
Figure 15 illustrates the system architecture based on distributed schemes synoptically.
In distributed schemes, in all websites, all be provided with first scheduler.Operation is submitted to first scheduler of each site-local, and dispatched by it.Because all websites all have first scheduler, therefore, about certain operation, in the site-local of regulation,,, then can re-execute scheduling, so that in these other site-locals, carry out this operation if other site-locals can have the free time by after the temporary transient scheduling.First scheduler of each site-local need have identical information, exchanges the information such as load state of each site-local at any time or termly.
Non-patent literature 1:Chris Smith, " Open Source Metascheduling forVirtual Organizations with Community Scheduler Framework (CSF) ", Technical Whitepaper, Platform Computing Inc.2003 August.
Non-patent literature 2:Vijay Subramani, " Distributed Job Scheduling onComputational Grids using Multiple Simultaneous Requests ", IEEEInternational Symposium on High Performance DistributedComputing (HPDC 2002), 2002.
Summary of the invention
As mentioned above, in the past,, the scheme of managing gridding set of systems integral body had been proposed in order to construct the distributed system of the wide area that fully utilizes the computer resource in a plurality of grid system groups that are present on the network.
But, above-mentioned concentrated scheme, since need be in first scheduler the details of existing computer resource in each site-local on the supervising the network, therefore, under the situation of appending change system architectures such as new grid system, for it is reflected in the scheduling based on first scheduler, the step of carrying out must be arranged in central server.Therefore, be accompanied by the change (change of the computer resource in the appending of site-local, deletion, the site-local etc.) of system architecture, ask the setting in the Singapore dollar scheduler more, thus increased use in the central server, the load of management.Therefore, the expansion of the change of answering system structure, scale is easily dwindled etc., thereby is difficult to become the system of variable size.
In addition, classification type scheme, from first scheduler of central server after the scheduler of site-local has been submitted operation to, unit's scheduler is not because directly related with this operation, therefore, even if behind submit job, in other site-locals, can there be the free time, the execution of each operation also is to carry out in each site-local that sends this operation.Thus, will not necessarily can carry out operation efficiently.
Have again, in classification type scheme, also can propose a kind of like this structure: by exchange message between each site-local, in other site-locals, can have under the idle situation, the operation of submitting to other site-locals is sent to empty site-local once more.But, when proposing such structure,, therefore under developmently need cost time because system can complicate.
Because therefore exchange message between each site-local, can increase offered load.In addition, even if in hierarchy plan, also identical with the concentrated scheme, owing to the change that is accompanied by system architecture needs setting in the Singapore dollar scheduler more, therefore, the expansion of the change of answering system structure, scale is easily dwindled, thereby is difficult to become the system of variable size.
In distributed schemes,, therefore, also can increase the load of network owing to must between first scheduler of all site-locals, exchange the information such as load state of each site-local at any time or termly.So, upgrade the setting in first scheduler of each site-local owing to the change that need be accompanied by system architecture, therefore, can not easily tackle change, the expansion of scale is dwindled, thereby be difficult to become the system of variable size in system architecture.
In addition, in classification type scheme and distributed schemes, owing to must develop first scheduler and local scheduler respectively, therefore, cost of development increases.
Therefore, the objective of the invention is to realize a kind of system architecture of size change: in the wide-area distribution type system of the computer resource in comprehensive utilization is present in the grid system group, the system architecture of the variable size that the expansion of the change of answering system structure, scale is easily dwindled.
Other purposes of the present invention are in the exchange message, alleviating offered load in order to carry out operation efficiently between site-local.
Another object of the present invention is not increase system development costs, just can realize combining the wide-area distribution type system of grid system group.
In order to achieve the above object, the present invention is embodied as the network that computer system group (grid system group) that utilization connects via network is carried out distributed treatment.That is, each grid system in this distributed processing environment has: grid service server is used for carrying out at the computer resource on the network distribution and the execution request of the operation of information processing; And, at the actual computer resources such as processing server (local resource) of carrying out operation in system.Each grid service server has: scheduler, and being used for the Resource Unit that comprises other grid systems on each local resource and the network is the distribution that object is carried out operation, and this operation is carried out in request; And proxy machine (Resource Broker machine) is used for communicating by letter between this scheduler of relaying and Resource Unit.This proxy machine is the software module that is used for the information of management resource unit, be used to have replaced distributing this Resource Unit of this operation to accept the execution request of the described operation that produces by scheduler, and, this Resource Unit is carried out the execution request of this operation according to the situation of this Resource Unit.
In more detail, proxy machine be on local resource and network with native system in abutting connection with each independent setting of other grid systems (Internet resources) of (directly being connected), each with corresponding Resource Unit between ask to carry out operation with the independent communication form that sets.
Proxy machine corresponding to local resource, obtain and manage the information relevant with operating conditions from relevant local resource with its ability, with the corresponding proxy machine of other grid systems (Internet resources), the grid service server from other relevant grid systems obtains and manages the available resource capability of information this grid service server is asked to(for) the execution of operation.Afterwards, each information that scheduler is managed according to the Resource Broker machine is carried out the distribution for the operation of Resource Unit.
In addition, this grid service server can have the interface unit that possesses resource capability information acquisition unit and resource capability information notice portion.Wherein, the resource capability information acquisition unit is replied the execution request from the operation of outside, obtains the information of the available resource capability of native system from described scheduler; The inquiry of the grid service server proxy machine in other grid systems on the automatic network is replied by described resource capability information notice portion, and the information of the available resource capability that resource capability information portion is accessed is notified to this proxy machine.In this case, scheduler calculates available resource capability according to the accessed described resource capability information of proxy machine from native system.Afterwards, give the resource capability information acquisition unit with the information of this available resource capability.
In addition, the interface unit of this grid service server can be set at the structure with operation receiving portion and operation execution trust portion.Wherein, operation receiving portion is accepted the execution request of the operation that the grid service server proxy machine other grid systems sends from network; Operation is carried out trust portion and is given scheduler with the operation relevant with the execution request that this operation receiving portion receives, and entrusts distribution and the execution of carrying out operation.
The present invention of other that achieve the above object also can be implemented as and is used for carrying out the scheduling of operation and carrying out the following operation execution control method of asking at grid system.This method comprises: computer system (grid system) is utilized the interface module with each corresponding setting of the interior local resource that is comprised of native system and other grid systems (Internet resources) on the network, from local resource, obtain and manage the ability of relevant local resource and the information of working condition, afterwards, obtain the also step of the information of the available resource capability of network resource administration in the grid service server from these Internet resources; According to these information, be the step that object is carried out the distribution of operation with the Resource Unit that comprises local resource and Internet resources by the interface module management; And, send the step of the request of carrying out at the operation of the Resource Unit that has distributed operation.In addition, also comprise the temporary transient operation that keeps being sent of this interface module and carry out request, and, send it to the step of related resource unit according to the working condition of the Resource Unit that has distributed operation.
In addition, the present invention also can be embodied as: control computer achieves the functional programs of above-mentioned grid service server or makes computing machine carry out the corresponding handling procedure of each step of carrying out control method with above-mentioned operation.This program or stored configuration perhaps provide via Web publishing in disk, CD, semiconductor memory, other recording mediums.
According to the present invention who as above constitutes, connect and compose the grid service server of each grid system of wide-area distribution type system via proxy machine, because it is by making the information of holding the grid system of serving as in this proxy machine, just can change all structures of grid system group accordingly by appending, delete the proxy machine of serving as again, therefore, can realize the system architecture of the variable size that the expansion of change, the scale of corresponding system architecture is easily dwindled.
According to the present invention, owing to the information of holding other grid systems of adjacency on network because of each grid system, the result makes all information of grid system group common for each grid system, therefore, message exchange can be between each grid system, carried out continually, thereby offered load can be alleviated.
In addition, according to the present invention, owing to do not need to be used for the intrinsic structure of Comprehensive Control grid system group integral body, therefore, can realize significantly cutting down required time of system development and cost etc., and this class of constructing the system with diverse network structure easily is rich in the system of dirigibility.
Description of drawings
Fig. 1 illustrates all structures according to the wide-area distribution type system of present embodiment.
Fig. 2 illustrates the structure of each grid system of the wide-area distribution type system of pie graph 1.
Fig. 3 schematically illustrates the example of hardware configuration of the computer installation of the grid service server that is suitable for realizing the present embodiment and processing server.
Fig. 4 illustrates the functional structure of the grid service server in the present embodiment.
Fig. 5 illustrates the relation of the Resource Broker machine of the functional structure of the processing server in the present embodiment and grid service server.
Fig. 6 illustrates the relation that wide-area distribution type system to present embodiment sends the scheduler of the functional structure of the client computer that operation carries out request and grid service server.
Fig. 7 is the process flow diagram that is used to illustrate according to the scheduling operation of the performed operation of the scheduler of present embodiment.
Fig. 8 illustrates the relation of Resource Broker machine, GS proxy machine interface and scheduler in the present embodiment.
Fig. 9 illustrates all structure of formation based on the grid system group of the wide-area distribution type system of present embodiment.
Figure 10 illustrates the distribution form under the situation that has dropped into operation in the grid system group that network plan connected by present embodiment in the grid system of stipulating.
Figure 11 illustrates in the grid system group of Figure 10, has dropped into the distribution form under the situation of operation at other grid systems.
Figure 12 is used to illustrate in the grid system group of Figure 10, dropped into the figure of the resource capability under the situation of operation to the grid system of regulation.
Figure 13 summarily illustrates the system architecture based on the wide-area distribution type system of concentrated scheme.
Figure 14 summarily illustrates the system architecture based on the wide-area distribution type system of classification type scheme.
Figure 15 summarily illustrates the system architecture based on the wide-area distribution type system of distributed schemes.
Embodiment
Below, with reference to the accompanying drawings, detailed description is used to implement best mode of the present invention (hereinafter referred to as embodiment).
Fig. 1 illustrates all structures based on the wide-area distribution type system of present embodiment.
The grid system group that the wide-area distribution type system of present embodiment will be connected on the wide area networks such as internet integrates, and can reciprocally utilize the computer resource in each grid system.Each grid system is a kind of by means of grid computing technology, and a plurality of diversified computer resource that is connected on the network is carried out management as 1 group, and its load disperses and the distributed processing system(DPS) of scheduling thereby carry out.
In the present embodiment, each grid system does not have subordinate relation, but with peer-to-peer executable operations concurrently.In addition, concerning each grid system, other grid systems of adjacency can be carried out processing in the same manner with the local resource (computer resource) in the native system on network, entrust and carry out operation.Here, so-called on network adjacency, mean and can directly carry out relation between the grid system of exchanges data via network.In addition, so-called local resource is meant the actual computer resources such as processing server of carrying out the operation that is distributed in grid computing.Below, the organizational form of the grid system group that will form based on this present embodiment is called network plan (Network Scheme).
Fig. 2 illustrates the structure of each grid system of the wide-area distribution type system of pie graph 1.
As shown in Figure 2, the grid system according to present embodiment has: the grid service server (GS) 100 of carrying out job assignment (scheduler); And, as according to based on the distribution of grid service server 100 and the actual processing server (PS) 200 of carrying out the local resource of operation.In addition, grid service server 100 also is connected with the grid service server 100 of other grid systems.
In the present embodiment, the grid service server 100 of grid service server 100 and processing server 200 and a plurality of grid systems they be connected with other computer networks by the internet each other.This computer network also can be regardless of communication protocol or wired or wireless communication form, and fire wall or other restrict access are set.
Will be described in the back in more detail, in order to realize above-mentioned network plan, the grid service server 100 of present embodiment has the interface module that is called proxy machine, and it is connected via grid service server 100 of this proxy machine and processing server 200, other grid systems etc.Below, the structure that connects between relevant apparatus is called proxy construction (Agent Framework).
Fig. 3 schematically illustrates the example of hardware configuration of the computer installation of the grid service server 100 that is suitable for realizing present embodiment and processing server 200.
Computer installation shown in Figure 3 has: as CPU (the CentralProcessing Unit: central processing unit) 11 of arithmetic element; Primary memory 13, it is via M/B (motherboard: mother board) chipset (chip set) 12 and cpu bus and be connected on the CPU 11; Video card 14, it is connected on the CPU 11 via this M/B chipset 12 and AGP (Accelerated GraphicsPort); Disk set (HDD) 15, network interface 16, they are connected on the M/B chipset 12 via PCI (Peripheral Component Interconnect) bus; In addition, also have floppy disk 18 and keyboard/mouse 19, they from this pci bus via low speed bus such as bridgt circuit 17 and ISA (Industry Standard Architecture) buses and be connected on the M/B chipset 12.
Fig. 3 only illustration realize the hardware configuration of the computer installation of present embodiment, if can use present embodiment, also can obtain other various structures.For example, replace video card 14 is set, both can constitute video memory only is installed, utilize CPU 11 to come the structure of image data processing; Can be used as external memory again, the driver of CD-R (Compact Disc Recordable) and DVD-RAM (Digital Versatile DiscRandom Access Memory) is set via ATA (AT Attachment) and SCSI interfaces such as (Small Computer System Interface).
Fig. 4 illustrates the functional structure of the grid service server 100 in the present embodiment.
Resource capability acquisition unit 112 is to each processing server 200 and the Resource Broker machine 120 inquiry native systems spendable resource capability corresponding with other grid systems of adjacency on network, and obtain information (below, with being identified as the processing server 200 of local resource of native system and other grid systems of being identified as Internet resources altogether, be called Resource Unit).In the information of obtaining, include the static information such as memory capacity of original processing ability and memory storage in the native system and based on real-time multidate informations such as load state.
Best resource selection portion 114 is according to by the accessed resource capability information of resource capability acquisition unit 112, selects to be suitable for most the Resource Unit of its execution according to operation, afterwards, carries out the distribution of operation.Optimized logic in this job assignment can be arbitrarily.
In the present embodiment, relaying is carried out in communicating by letter between 120 pairs of schedulers 110 of Resource Broker machine and spendable Resource Unit, and replaces these request units to accept the execution request of the operation made by scheduler 110.Thus, though the object that sends of the request of the inquiry object of resource capability acquisition unit 112, operation trust portion 115 becomes Resource Broker machine 120, the function of scheduler 110 in addition itself does not change with scheduler in the past.The difference of the communication form between the grid service server 100 of grid service server 100 and each processing server 200 or other grid systems, utilize the setting in the Resource Broker machine 120 and be absorbed, therefore, do not need to consider the difference of scheduler 110 from the communication form when sending request.Thus, scheduler 110 can use at the employed scheduler of grid system in the past.
The corresponding Resource Unit of resource situation management department 121 visits is to hold current operating conditions in this processing server 200 (local resource) or the grid system (Internet resources).
The statistical information that resource capability management department 122 management is relevant with the operation executive capability of corresponding Resource Unit etc., and, return institute's information of managing according to inquiry from the resource capability acquisition unit 112 of scheduler 110.Here, the statistical information of so-called relevant operation executive capability etc. is not only the static information of the memory capacity of the processing power of CPU self and memory storage, and the time that also comprises at the load of CPU changes and the resulting information of this class dynamic content execution statistical treatment of tendency of operation.
By this resource information that resource situation management department 121 and resource capability management department 122 are managed, obtain from corresponding Resource Unit by Resource Broker machine 120, and it is stored in the storage unit such as primary memory 13, disk set 15 of Fig. 3.
GS proxy machine interface 130 is realized by be subjected to storage unit such as programme controlled CPU 11, primary memory 13 or disk set 15 and network interface 16 for example shown in Figure 3, as its concrete function, as shown in Figure 4, it has resource capability information acquisition unit 131, resource capability information notice portion 132, operation receiving portion 133 and operation execution trust portion 134.
As mentioned above, GS proxy machine interface 130 has a kind of like this function, and it is used to make grid system with respect to other grid systems on the network, can similarly use with local resource that mutually should other grid systems.Utilize this GS proxy machine interface 130, grid service server 100 can be accepted the request from the grid service server 100 of other grid systems, and returns and use the spendable Resource Unit of native system to execute the result of operation gained.
Resource capability information acquisition unit 131 is to scheduler 110 inquiry and obtain at the resource capability information (resource information) of carrying out the available native system of request from the operation of outside.
Resource capability information notice portion 132 obtains request according to the resource capability that receives, and the resource information that will get access in resource capability information acquisition unit 131 is notified to the transmission source that resource capability obtains request.The Resource Broker machine 120 of the correspondence of the grid service server 100 other grid systems from network has received resource capability and has obtained under the situation of request, and it is notified to this Resource Broker machine 120.In Resource Broker machine 120, resource situation management department 121 and resource capability management department 122 accept this notice, and are stored in them in memory storages such as primary memory 13 or disk set 15 and the execution management.Issue the notice of grid service server 100 from resource capability information notice portion 132, both can carry out termly, carry out in the time of also can change taking place in the working condition of native system.In addition, also can be at any time, send inquiry from the Resource Broker machine 120 of grid service server 100 to the grid service server 100 of other grid systems of correspondence.
Operation is carried out the scheduling and the execution of the operation that trust portion 134 will be accepted by operation receiving portion 133 and is entrusted to scheduler 110.
Next, the relation between the functional structure of processing server 200 and the corresponding Resource Broker machine 120 is described.
Fig. 5 illustrates the relation between the Resource Broker machine 120 of the functional structure of processing server 200 and grid service server 100.
As shown in Figure 5, processing server 200 has processing server with Resource Broker machine interface (below, be called " PS proxy machine interface ") 210, is used for making computer installation shown in Figure 3 processing server 200 as grid system.
PS proxy machine interface 210 is realized by be subjected to storage unit such as programme controlled CPU 11, storer 13 or disk set 15 and network interface 16 for example shown in Figure 3, as its concrete function, as shown in Figure 5, it has PS condition monitoring portion 211, resource capability information notice portion 212, operation receiving portion 213 and operation execution portion 214.
Current behaviour in service, the resource situation of PS condition monitoring portion 211 these devices of supervision (processing server 200) are come acquisition of information.
Resource capability information notice portion 212 is notified to the information relevant with resource situation with behaviour in service PS that collect in the PS condition monitoring portion 211 in the Resource Broker machine 120 of grid service server 100.In Resource Broker machine 120, resource situation management department 121 and resource capability management department 122 accept this notice, and are stored in them in memory storages such as primary memory 13 or disk set 15 and the execution management.The notice that sends to grid service server 100 from resource capability information notice portion 212 both can be carried out termly, carried out in the time of also can change taking place in the working condition of processing server 200.In addition, also can be at any time, send inquiry from the Resource Broker machine 120 of the correspondence of the webserver 100 to processing server 200.
Operation receiving portion 213 is accepted to carry out request from the operation that the operation trust portion 124 of the Resource Broker machine 120 of grid service server 100 sends out.
Operation execution portion 214 uses the resource of processing server 200 to carry out the operation of being accepted by operation receiving portion 213.
If the GS proxy machine interface 130 of grid service server 100 and the PS proxy machine interface 210 of processing server 200 are compared, following some difference is arranged: the situation of PS condition monitoring portion 211 these devices of supervision is come acquisition of information, relative therewith, resource capability information acquisition unit 131 is to the resource capability of scheduler 110 inquiry native systems; In addition, operation execution portion 214 uses the resource of this device to carry out operation, and is relative therewith, and trust portion 134 entrusts execution from operation to scheduler 110 is carried out in operation.This is based on following situation and different: the processing server 200 that PS proxy machine interface 210 is housed is local resources of carrying out operation in grid system, relative therewith, the grid service server 100 that GS proxy machine interface 130 has been installed is Comprehensive Control grid systems, and carries out the server of the scheduling of operation.
Promptly, in the Resource Broker machine 120 of the grid service server 100 that request that resource capability is obtained or the execution request of operation send and corresponding GS proxy machine interface 130 and the relation between the PS proxy machine interface 210, GS proxy machine interface 130 and PS proxy machine interface 210 are without any different.Therefore, Resource Broker machine 120 all has the identical functions structure no matter the corresponding processing server 200 to liking as local resource still is as other grid service servers 100 of Internet resources.
So, because Resource Broker machine 120 is used as interface module, and in grid service server 100, has GS proxy machine interface 130, therefore, between the grid service server 100 of adjacency on the network, can reciprocally handle other system in the same manner with the local resource of native system, and carry out the distribution of operation, and can move as the local resource of other system.Utilize this proxy machine structure, realized carrying out following network operating scheme: do not have subordinate relation between the grid system and with equity concern executed in parallel operation, other grid systems can be handled as the resource of native system each other simultaneously.
Next, just describe based on the operation of the wide-area distribution type system of the present embodiment that constitutes as mentioned above.
In the present embodiment, carry out request to the operation that the grid system of stipulating sends, utilize the grid system group that constitutes the wide-area distribution type system to carry out the load dispersion, thereby carry out operation according to client computer from regulation.Here, so-called client computer is computing machine, the PDA information equipments such as (Personal DigitalAssistant) of any one grid system of the wide-area distribution type system of addressable formation present embodiment.Processing server 200 with aftermentioned function also can be used as client computer and sends operation execution request.
Fig. 6 illustrates the relation between the scheduler 110 that wide-area distribution type system to present embodiment sends the functional structure of the client computer that operation carries out request and grid service server 100.
As shown in Figure 6, client computer 300 has and is used to inquire the resource capability inquiry portion 310 of grid system resource ability that carries out the transmission object of request as operation, and, send operation and carry out request, and send it to the operation trust portion 320 of grid system.Client computer 300 can obtain the execution result at desired operation, and the supply of necessary computer resource also can be entrusted under the situation of grid system in the execution of operation, and resource capability inquiry portion 310 is constitutive requirements not necessarily.These functions for example are being under client computer 300 situation about being made of computer installation shown in Figure 3, by realized by programme controlled CPU 11 and storage unit such as primary memory 13 or disk set 15.
When carrying out operation according to present embodiment, at first, the operation trust portion 320 of above-mentioned client computer 300 sends operation and carries out request, and sends it to the grid service server 100 in the grid system of access object.Here,, send resource capability from resource capability inquiry portion 310 to same grid service server 100 and obtain request, whether have the resource capability that the execution operation is satisfied so that can judge grid system as prior investigation before this.
In the scheduler 110 of grid service server 100, operation receiving portion 113 is accepted to carry out request from the operation that client computer 300 is sent, and best resource selection portion 114 is an object with the spendable Resource Unit of native system, carries out the distribution of this operation.As the Resource Unit of the distribution object of operation is as the processing server 200 of local resource and as other grid systems of Internet resources.
Fig. 7 is the process flow diagram that is used to illustrate the scheduling operation of the operation that is produced by scheduler 110.
With reference to Fig. 7, best resource selection portion 114 is via resource capability inquire response portion 111 and resource capability acquisition unit 112, obtain the ability, tendency of operation etc. (steps 701) such as statistical informations of each Resource Unit from Resource Broker machine 120, afterwards, carry out optimal scheduling (step 702) according to kind, the characteristic of these information and operation.Afterwards, operation trust portion 115 is according to the result based on best resource selection portion 114, just send operation execution request no matter be assigned with the working condition of the Resource Unit of operation, and send it to the Resource Broker machine 120 (step 703) corresponding with this Resource Unit.
The logic of the scheduling of being carried out by best resource selection portion 114 can be arbitrarily, but because if entrust the execution operation to other grid systems, then both be as the grid system of entrusting object, also can carry out the scheduling that operation is carried out, therefore, generally we think the operating efficiency height that distributes operation to local resource.Therefore, can adopt following method: at first distribute operations, only in processing server 200, under the situation of scarce capacity, just other grid systems are entrusted in the execution of operation to processing server 200 as the local resource of native system.
The execution result of the operation that scheduler 110 will obtain based on each Resource Unit of collecting from Resource Broker machine 120 integrates, and it is returned to client computer 300.
And in the present embodiment, the Resource Unit of execution operation has the situation as the processing server 200 of the local resource of the grid system of on commission execution operation, also has the situation as other grid systems of Internet resources.Wherein, in processing server 200, having accepted operation from the Resource Broker machine 120 of grid service server 100 when the operation receiving portion 213 of PS proxy machine interface 210 carries out under the situation of request, its operation execution portion 214 asks to carry out operation according to this, and execution result is turned back to the Resource Broker machine 120 of grid service server 100.
On the other hand, at Resource Unit is under the situation of grid system, carry out request from the operation of Resource Broker machine 120 and accept, carry out the scheduler 110 that trust portion 134 sends to this grid service server 100 by operation by the operation receiving portion 133 of the GS proxy machine interface 130 in the grid service server 100 of this grid system.
Fig. 8 illustrates the relation between Resource Broker machine 120, GS proxy machine interface 130 and the scheduler 110.
Here, comparison diagram 8 and Fig. 6, for scheduler 110, with the relation of GS proxy machine interface 130 and with the relation of the PS proxy machine interface 210 of client computer 300 be of equal value.Therefore, scheduler 110 with carry out at the operation of directly collecting from above-mentioned client computer 300 requested operation in the same manner, carry out request at the operation of collecting from the grid service server 100 of other grid systems and come operation dispatching via GS proxy machine interface 130, thereby the execution of operation can be entrusted to the spendable Resource Unit of native system.
But Resource Broker machine 120 obtains the information (resource information) relevant with the operation executive capability with the work at present situation as described above from the Resource Unit of correspondence, and utilizes resource situation management department 121 and resource capability management department 122 to carry out management.At Resource Unit is under the situation of processing server 200, and this resource information is collected by the PS condition monitoring portion 211 of its PS proxy machine interface 210, afterwards, sends it to Resource Broker machine 120 by resource capability information notice portion 212.
On the other hand, at Resource Unit is under the situation of other grid systems, resource information is obtained by the resource capability information acquisition unit 131 of the GS proxy machine interface 130 in the grid service server 100 of this grid system, and sends to Resource Broker machine 120 by resource capability information notice portion 132.Here, resource capability information acquisition unit 131 is inquired to the resource capability inquire response portion 111 of scheduler 110, and is accepted these information as shown in Figure 4.Therefore, there is the situation that receives the inquiry of resource capability from client computer 300 in resource capability inquire response portion 111, and the situation of accepting inquiry from GS proxy machine interface 130 is also arranged.
Under the situation that is conceived to 1 grid system, other grid systems with this grid system adjacency on network can use as the Resource Unit of this grid system, on the other hand, also can entrust at this grid system and carry out operation.Then, if consider to entrust the situation of carrying out operation from the grid system of regulation to this grid system, then this grid system can not use the grid system of entrusting the regulation of the carrying out operation Resource Unit as native system.
Therefore, under the situation of having accepted inquiry from GS proxy machine interface 130, scheduler 110 can not use the grid system that comprises a kind of like this grid service server 100 as Resource Unit, wherein, described grid service server 100 is that GS proxy machine interface 130 was sent the grid service server that resource capability obtains request.Therefore, in this case, scheduler 110 will be removed to comprise and send the grid system that resource capability obtains the grid service server 100 of request and calculate available resource capability, and it is returned to GS proxy machine interface 130.
Fig. 9 illustrates the one-piece construction of formation based on the grid system group of the wide-area distribution type system of present embodiment.
As mentioned above, in the present embodiment, in each grid system, grid service server 100 with as the processing server 200 of local resource be connected and being connected of grid service server 100 and other grid systems is to carry out via the Resource Broker machines 120 that had in the grid service server 100.Thus, realized network plan shown in Figure 9, each grid system group can be accepted the execution trust of operation respectively from client computer 300.This operation can be carried out by the processing server 200 as the native system local resource, perhaps also can put on other grid systems of adjacency on the network and carry out.Do not have subordinate relation between each grid system but operate with the peer-to-peer executed in parallel.
Figure 10 illustrates the distribution form under the situation that grid system to the regulation of the grid system group that network plan connected by present embodiment dropped into operation.
In the example of Figure 10, the grid A in the wide-area distribution type system that is made of 5 grid systems (grid A, B, C, D, E) shown in the dotted line has been dropped into operation.This operation is at first disperseed to put in the processing server (PS) 200 as the local resource of grid A.Afterwards, be untreated this operation and under the situation that taken place to overflow in ability with the local resource of grid A, in the grid service server (GS) 100 of grid A, and network between the corresponding Resource Broker machine 120 of grid B, C and scheduler 110 of adjacency, the negotiations such as affirmation of execution work situation, and this operation put into grid B, C.Operation is being put under the situation of grid B, C, utilized the local resource of each grid B, C to carry out operation, under the situation that in grid B, has taken place to overflow, also this operation can disperseed to put among grid D, the E with grid B adjacency.
According to the kind of using, can transmit operation from grid A to grid B, C regardless of the working condition of grid B, C etc.In this case, in the grid service server 100 of grid A, the execution of having accepted this operation at the Resource Broker machine 120 corresponding with grid B, C entrust and grid B, C become the state that can accept operation the time engrave, entrust to the execution that grid B, C carry out operation from Resource Broker machine 120.
In the above description, be set in the local resource of native system, under the situation that has taken place to overflow not handling operation, operation put in other grid systems.Utilizing local resource to come the method for priority processing operation so, as far as possible is preferred in the load that alleviates network.But the dispersion input method of operation is not limited to this.Can also carry out the distribution of operation with logic arbitrarily, so that can carry out best (it is high to carry out efficient) dispersion according to the kind of the ability of other grid systems of the local resource of native system and adjacency, operation, characteristic etc.
Figure 11 illustrates in the grid system group of Figure 10, the distribution form under the situation of the operation that drops at other grid systems (grid B).
As shown in figure 11, at first be put in the processing server 200 as the local resource of grid B drop into for the operation of grid B, under the situation of overflowing, its dispersion put among grid A, D, the E of adjacency on network.If in grid A, produced and overflowed, then also operation can be disperseed to put among the grid C.
So, in utilizing the grid system group that network plan connected of present embodiment, since between each grid system dereliction from the relation but operate with the peer-to-peer executed in parallel, therefore, both just operation had been dropped into to any one grid system, also can disperse to drop into operation to the grid system of adjacency on network chainly, thus can be in the wide-area distribution type entire system that constitutes by the grid system group processing operation.
Here, only just carry out under the situation of operation according to the wide-area distribution type system of present embodiment, an example of the computing method of resource capability describes.
Figure 12 is the grid system group that is used for illustrating Figure 10, the figure of the resource capability under the situation that has dropped into operation to the grid system of stipulating.
In Figure 12, the resource capability of each grid system is as giving a definition.
C
x: (local resource) resource capability of grid system x self
C
X for the user: carry out request, the available resource capability of grid system x at operation from client computer;
C
X for y: at the execution request that comes on the comfortable network with the operation of the grid system y of grid system x adjacency, the available resource capability of grid system x, that is, and C
X for the userAnd C
X for yCalculate by following formula 1.
[formula 1]
C
X for the user=C
x+ ∑
The all of its neighbor point of y ∈ xC
Y for x
C
X for y=C
x+ ∑
Z ∈ all of its neighbor point except that y, xC
Z for x
Referring to Figure 12, under the situation of the execution request of having sent operation from client computer 300 to grid system a (the grid A of Figure 10), can following calculating be used to carry out this operation and the processing power C that provides
A for the user
C
A for the user=C
a+ C
B for a+ C
C for a
Here, since grid system b (the grid B among Figure 10) except with grid system a in abutting connection with also with grid system d, e (grid D, the E of Figure 10) adjacency, therefore, the resource capability that can provide to grid system a adds the resource capability that provides from grid system d, e for the resource capability of native system.That is, for
C
B for a=C
b+ C
D for b+ C
E for b
On the other hand and since grid system c (the grid C of Figure 10) only with grid system a adjacency, therefore, only can provide the resource capability of native system, thereby become
C
C for a=C
c
Equally, according to Figure 12, under the situation of execution request from client computer 300 to grid system b that carried out operation from, the processing power C that provides in order to carry out this operation
B for the userCan followingly calculate.
C
B for the user=C
b+ C
A for b+ C
D for b+ C
E for b
Like this, it is as follows to offer grid system b resource capability respectively from grid system a, d, e:
C
A for b=C
a+ C
C for a
C
D for b=C
d
C
E for b=C
e
Under the situation of execution request from client computer 300 to grid system d that carried out operation from, for carrying out the processing power C that this operation provides
D for the userFollowing calculating.
C
D for the user=C
d+ C
B for d
Like this, it is as follows to offer the resource capability of grid system d from grid system b:
C
B for d=C
b+ C
A for b+ C
E for b
Wherein, resource capability C
A for b, C
E for bBe such as mentioned above.
In addition, above computing method are illustration only, is not limited to this method certainly.
As mentioned above, in the present embodiment, each grid system by directly asking for native system resource capability and with resource capability that other grid systems of native system adjacency provide, can know that having carried out operation at the grid system of regulation carries out all processing poweies of wide-area distribution type system under the situation of request.Therefore, do not need conventional art shown in Figure 13-15 that the message exchange that first scheduler is carried out the state that is used to hold all grid systems of wide-area distribution type system and its local resource etc. is set like that, thereby can alleviate offered load significantly.
Present embodiment is that interface module is connected with carrying out as the processing server 200 of local resource the grid service server 100 of grid system with the Resource Broker machine 120 that is provided with in the grid service server 100, simultaneously, connect grid service server 100 via same Resource Broker machine 120 and the grid service server 100 of other grid systems of adjacency on network.Thus, the grid service server 100 of each grid system of adjacency can be handled other grid system each other in the same manner with the local resource of native system on network, thus, can realize utilizing above-mentioned network plan to connect the wide-area distribution type system that the grid system group forms.
Therefore the scheduler 110 of each grid service server 100, there is no need to import the special construction that the wide-area distribution type system uses owing to do not need to distinguish the local resource of other grid systems and native system.In addition, also there is no need to be provided with first scheduler that the grid system group execution that constitutes the wide-area distribution type system is synthetically managed.Thus, can significantly cut down required time of system development and cost etc.
In addition, in the present embodiment, because the information of the Resource Unit that is made of other grid systems of local resource and adjacency is by with Resource Broker machine 120 management of the corresponding setting of each Resource Unit, therefore, scheduler 110 also can not considered the duty of each Resource Unit, and Resource Broker machine 120 is carried out job assignment.Therefore, or appended new grid system or from the grid system group, removed under the situation of grid system of regulation to the grid system that constitutes the wide-area distribution type system, with the grid system of these grid system adjacency in, can carry out correspondence by only appending or delete corresponding Resource Broker machine 120.Thus, the extendability of system, dirigibility etc. will be very high.
Like this, serve as the number of Resource Broker machine 120 of the grid system of adjacency on network by plus-minus, can easily construct the wide-area distribution type system that hierarchy construction and cascade construction etc. have the arbitrary network structure.
For example, by a plurality of Resource Broker machines 120 of serving as other grid systems only are set in 1 grid system, can construct this 1 grid system is used as backup application mode system that use, that be similar to the concentrated scheme as the center with other grid system groups.
In addition, in each grid system,, can construct the system that cascade connects each grid system by 1 Resource Broker machine 120 of serving as other grid systems only is set.
In addition, according to the setting of Resource Broker machine 120,, therefore, can construct firm wide-area distribution type system at fault owing to can easily be defined in employed alternative grid system under the situation that fault has taken place on the network.
Claims (14)
1. a computer system that is connected with this network in the distributed processing environment that connects a plurality of computer systems via network is characterized in that, comprising:
Carry out the computer resource of operation, i.e. local resource; And
The grid service server of carrying out job assignment and sending the request of execution,
Wherein, described grid service server has:
Proxy machine, management with on described network, be connected to described computer system, be on the native system at least 1 other computer system, be the relevant information of Internet resources and described local resource, be resource information, and ask these resources execution operation; And
Scheduler, acknowledges requests are carried out the 1st request of the 1st operation, based on described resource information, give one or more resources in described local resource and the described Internet resources with the 1st job assignment.
2. computer system as claimed in claim 1 is characterized in that, described proxy machine be in described Internet resources and the described local resource each and be provided with one by one.
3. computer system as claimed in claim 2 is characterized in that, the described resource information of being managed by described proxy machine comprises the static information relevant with the processing power of pairing described local resource or described Internet resources.
4. computer system as claimed in claim 2 is characterized in that, the described resource information of being managed by described proxy machine comprises the multidate information relevant with the real-time load state of pairing described local resource or described Internet resources.
5. computer system as claimed in claim 1, it is characterized in that, described grid service server also has the 1st interface unit, the 2nd request of the 2nd operation is carried out in the request that the acceptance of the 1st interface unit is sent from other computer systems on the described network, the 2nd request is sent to described scheduler, and entrusts the distribution of the 2nd operation to it.
6. computer system as claimed in claim 5, it is characterized in that, described the 1st interface unit also have response from the inquiry of described other computer systems and with relevant described computer system, be the unit that the described resource information of native system is notified to these other computer systems.
7. computer system as claimed in claim 2 is characterized in that,
The information will be with the processing power of this local resource relevant with working condition with the corresponding described proxy machine of described local resource manages as described resource information;
The described proxy machine corresponding with described Internet resources will manage as described resource information at the relevant information of the processing power of the available resource of described the 1st request with these Internet resources;
The described resource information that described scheduler is managed based on described proxy machine is carried out the distribution of described the 1st operation.
8. computer system as claimed in claim 2 is characterized in that described local resource also has the 2nd interface unit, and the response of the 2nd interface unit is notified to this proxy machine from the inquiry of pairing described proxy machine with described resource information.
One kind be connected with via network a plurality ofly have the local computer resource of carrying out operation respectively, be that control method is carried out in the operation of the execution of control operation in the distributed processing environment of computer system of local resource, it is characterized in that, comprising:
The 1st computer system that is connected to described network respectively from self described local resource obtain 1st resource information relevant with its processing power and to its manage and from least one the 2nd computer system of on described network, being connected with the 1st computer system, be that Internet resources obtain 2nd resource information relevant with its processing power and to its step that manages;
Described the 1st computer system acknowledges requests is carried out the 1st request of the 1st operation, based on the described the 1st and the 2nd resource information, is the step that object is carried out the distribution of described the 1st operation with described local resource and described Internet resources; And
Described the 1st computer system is to being assigned with the step of carrying out the 2nd request of the 1st operation with the request of sending of the resource of described the 1st operation.
10. method as claimed in claim 9, it is characterized in that, described the 1st computer system also has described the 2nd request that temporary transient maintenance is sent, and responds the working condition that is assigned with the described resource of described the 1st operation, sends the step of the 2nd request to this resource.
11. method as claimed in claim 10, it is characterized in that, described the 2nd computer system according to the relation of at least 1 the 3rd computer system that on described network, is connected with the 2nd computer system, carry out and the equal step of described the 1st computer system.
12. method as claimed in claim 9 is characterized in that, also has described the 2nd computer system and replys inquiry from described the 1st computer system, and notify the step of described the 2nd resource information.
13. one kind connected via network a plurality ofly have the local computer resource of carrying out operation respectively, be in the distributed processing environment of computer system of local resource, make the 1st computer system that is connected to described network carry out the program of following processing, it is characterized in that described being treated to:
The 1st computer system that is connected to described network from the described local resource of self, obtain respectively about the 1st resource information of its processing power and to its manage, from least one the 2nd computer system of on described network, connecting with the 1st computer system, be that Internet resources obtain 2nd resource information relevant with its processing power and to its processing that manages;
Described the 1st computer system acknowledges requests is carried out the 1st request of the 1st operation, based on the described the 1st and the 2nd resource information, is the processing that object is carried out the distribution of described the 1st operation with described local resource and described Internet resources; And
Described the 1st computer system is to being assigned with the 2nd processing of request of carrying out the 1st operation with the request of sending of the resource of described the 1st operation.
14. program as claimed in claim 13, it is characterized in that, also make described computing machine carry out following the processing: described the 1st computer system is temporary transient to keep described the 2nd request of being sent, and response is assigned with the working condition with the described resource of described the 1st operation, sends it to this resource.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2004161819 | 2004-05-31 | ||
| JP161819/2004 | 2004-05-31 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1954295A true CN1954295A (en) | 2007-04-25 |
Family
ID=35451046
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2005800154954A Pending CN1954295A (en) | 2004-05-31 | 2005-05-23 | Computer system, method, and program for controlling execution of job in distributed processing environment |
Country Status (3)
| Country | Link |
|---|---|
| JP (1) | JPWO2005116832A1 (en) |
| CN (1) | CN1954295A (en) |
| WO (1) | WO2005116832A1 (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101350752B (en) * | 2007-07-20 | 2013-04-24 | 美国博通公司 | Method and system for processing management information in communication network |
| CN106899656A (en) * | 2017-01-03 | 2017-06-27 | 珠海格力电器股份有限公司 | Equipment control method and device |
| CN110032364A (en) * | 2019-04-11 | 2019-07-19 | 上海商汤智能科技有限公司 | Data processing method, device, electronic equipment and computer storage medium |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4522780B2 (en) * | 2004-07-28 | 2010-08-11 | 株式会社トヨタＩｔ開発センター | Grid computing system, program, recording medium, and grid computing method |
| JP4806362B2 (en) * | 2007-02-14 | 2011-11-02 | 富士通株式会社 | Parallel processing control program, parallel processing control system, and parallel processing control method |
| JP4821783B2 (en) * | 2008-02-08 | 2011-11-24 | 日本電気株式会社 | Grid computing system and data processing method |
| US8959525B2 (en) * | 2009-10-28 | 2015-02-17 | International Business Machines Corporation | Systems and methods for affinity driven distributed scheduling of parallel computations |
| JP2013239124A (en) * | 2012-05-17 | 2013-11-28 | Nec Corp | Terminal control system, terminal management device, terminal control device, terminal control method, terminal management program, and terminal control program |
| JP6413789B2 (en) | 2015-01-22 | 2018-10-31 | 富士通株式会社 | Job management program, job management method, and job management apparatus |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2580525B2 (en) * | 1993-11-17 | 1997-02-12 | 工業技術院長 | Load balancing method for parallel computers |
| JP3512080B2 (en) * | 1995-12-27 | 2004-03-29 | ソニー株式会社 | Computing device and method |
| JP3745820B2 (en) * | 1996-02-23 | 2006-02-15 | 三菱電機株式会社 | Autonomous cooperative information processing apparatus and autonomous cooperative information processing method |
| JP2912225B2 (en) * | 1996-04-18 | 1999-06-28 | 四国日本電気ソフトウェア株式会社 | Communication processing system |
-
2005
- 2005-05-23 WO PCT/JP2005/009350 patent/WO2005116832A1/en active Application Filing
- 2005-05-23 CN CNA2005800154954A patent/CN1954295A/en active Pending
- 2005-05-23 JP JP2006513869A patent/JPWO2005116832A1/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101350752B (en) * | 2007-07-20 | 2013-04-24 | 美国博通公司 | Method and system for processing management information in communication network |
| CN106899656A (en) * | 2017-01-03 | 2017-06-27 | 珠海格力电器股份有限公司 | Equipment control method and device |
| CN110032364A (en) * | 2019-04-11 | 2019-07-19 | 上海商汤智能科技有限公司 | Data processing method, device, electronic equipment and computer storage medium |
| CN110032364B (en) * | 2019-04-11 | 2023-08-15 | 上海商汤智能科技有限公司 | Data processing method, device, electronic equipment and computer storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2005116832A1 (en) | 2005-12-08 |
| JPWO2005116832A1 (en) | 2008-04-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP4954089B2 (en) | Method, system, and computer program for facilitating comprehensive grid environment management by monitoring and distributing grid activity | |
| CN1954295A (en) | Computer system, method, and program for controlling execution of job in distributed processing environment | |
| US7788375B2 (en) | Coordinating the monitoring, management, and prediction of unintended changes within a grid environment | |
| US7707288B2 (en) | Automatically building a locally managed virtual node grouping to handle a grid job requiring a degree of resource parallelism within a grid environment | |
| JP4304535B2 (en) | Information processing apparatus, program, modular system operation management system, and component selection method | |
| US9075659B2 (en) | Task allocation in a computer network | |
| EP1408407A2 (en) | Method and apparatus for task allocation in a distributed computing environment | |
| CN102591921A (en) | Scheduling and management in a personal datacenter | |
| US9424096B2 (en) | Task allocation in a computer network | |
| CN111464659A (en) | Node scheduling method, node pre-selection processing method, device, equipment and medium | |
| US20050273511A1 (en) | Equitable resource sharing in grid-based computing environments | |
| JP2006018561A (en) | Method and program for allocating resource | |
| CN111506297A (en) | Mobile top layer design system | |
| Barlaskar et al. | Energy-efficient virtual machine placement using enhanced firefly algorithm | |
| In et al. | Sphinx: A scheduling middleware for data intensive applications on a grid | |
| Hao et al. | Evaluation of nine heuristic algorithms with data‐intensive jobs and computing‐intensive jobs in a dynamic environment | |
| Liu et al. | A Distributed Resource Sharing Mechanism in Edge-Enabled IIoT Systems | |
| CN101390056A (en) | Application system intelligent optimizer | |
| CN115237548A (en) | Multi-machine cooperation equipment scheduling method and device, electronic equipment and storage medium | |
| CN100358289C (en) | Comprehensive network managing system and method for managing network system | |
| CN114090256A (en) | Application delivery load management method and system based on cloud computing | |
| CN102611578A (en) | Network equipment data management system in multi-network-equipment environment | |
| JPH10207847A (en) | Automatic load dispersion system for distribution system | |
| CN112148462A (en) | Jenkins-based CICD process processing method | |
| CN118118488B (en) | Cloud data service dynamic self-adaptive adjustment method and system based on cloud edge cooperation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |