WO2018099090A1 - 从云计算系统中确定主调度器的方法及装置 - Google Patents

从云计算系统中确定主调度器的方法及装置 Download PDF

Info

Publication number
WO2018099090A1
WO2018099090A1 PCT/CN2017/092854 CN2017092854W WO2018099090A1 WO 2018099090 A1 WO2018099090 A1 WO 2018099090A1 CN 2017092854 W CN2017092854 W CN 2017092854W WO 2018099090 A1 WO2018099090 A1 WO 2018099090A1
Authority
WO
WIPO (PCT)
Prior art keywords
scheduler
computing system
cloud computing
schedulers
voting information
Prior art date
Application number
PCT/CN2017/092854
Other languages
English (en)
French (fr)
Inventor
刘力力
Original Assignee
华为技术有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Priority to EP17877361.0A priority Critical patent/EP3541048A4/en
Publication of WO2018099090A1 publication Critical patent/WO2018099090A1/zh
Priority to US16/425,680 priority patent/US20190280945A1/en

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/30—Decision processes by autonomous network management units using voting and bidding
    • 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
    • G06F9/465—Distributed object oriented systems
    • 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]
    • G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • G06F9/5038—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration
    • 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]
    • G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • G06F9/505—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the load
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L65/00—Network arrangements, protocols or services for supporting real-time applications in data packet communication
    • H04L65/40—Support for services or applications
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
    • H04L67/1004—Server selection for load balancing
    • H04L67/1008—Server selection for load balancing based on parameters of servers, e.g. available memory or workload
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/50—Network services
    • H04L67/60—Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources

Definitions

  • the present invention relates to the field of cloud computing technologies, and in particular, to a method and apparatus for determining a master scheduler from a cloud computing system.
  • cloud computing is a computing model.
  • the whole cloud computing system is mainly composed of a client, a scheduler and a resource pool.
  • the scheduler can receive The request of the client, and according to the request of the client, schedules the computing resources in the resource pool for use by the client.
  • each scheduler corresponds to a scheduling policy and is responsible for one type of service; for example, a cloud computing system Carrying three types of services, namely search engine, video surveillance, and big data; then, as shown in Figure 2, there may be three schedulers in the entire cloud computing system, namely a search engine scheduler, a video monitoring scheduler, and The big data scheduler; correspondingly, the search engine scheduler corresponds to the search engine's scheduling policy, and is responsible for managing the search engine client's business, that is, the search engine scheduler receives the search engine client's service request, and schedules the computing resources in the resource pool to search.
  • the search engine scheduler allocates the computing resource M in the resource pool to the search engine client when receiving the service request of the search engine client;
  • the engine scheduler and the big data scheduler are independent of each other, and may appear.
  • the big data server receives the service request of the big data client, the computing resource M in the resource pool is also allocated to the big data client, thereby Causes conflicts in the allocated computing resources.
  • Embodiments of the present invention provide a method and apparatus for determining a primary scheduler from a cloud computing system to avoid conflicts of allocated computing resources between different schedulers.
  • a method for determining a master scheduler from a cloud computing system includes at least an arbiter and a plurality of schedulers, each scheduler corresponding to a scheduling policy, the method comprising: the arbiter receiving at least one The voting information sent by the scheduler, the voting information is the scheduler according to the current load status of the cloud computing system and the preset negative
  • the voting information is generated by the current scheduler being the voting information of at least one of the plurality of schedulers, and the voting information may be specifically the identifier of the voted scheduler.
  • the arbiter determines the master scheduler from the plurality of schedulers according to the vote information sent by the at least one scheduler, so that the master scheduler is in the cloud computing system according to the corresponding scheduling policy.
  • the computing resources are scheduled, wherein the scheduling policy corresponding to the primary scheduler matches the current load status of the cloud computing system.
  • the arbiter selects a scheduler from the plurality of schedulers as the primary scheduler according to the voting information sent by the at least one scheduler, including: the arbiter according to the at least one schedule
  • the voting information sent by the device determines the voting score of each scheduler; the arbiter determines, from the plurality of schedulers, a scheduler whose voting score satisfies a preset condition as the primary scheduler.
  • the scheduler that meets the preset condition may be selected as the main scheduler, and the master scheduler may schedule the entire cloud computing system according to the corresponding scheduling policy, and the user may set different presets. The condition, and then the different schedulers are selected as the master scheduler, so that in the embodiment of the present invention, the user can flexibly select the master scheduler according to requirements.
  • a second possible implementation manner different priorities are set in advance for different schedulers, and the arbiter determines each scheduler according to the voting information sent by the at least one scheduler.
  • the voting score includes: the arbiter sets different weights for each scheduler's voting information according to the priority of each scheduler; the arbiter according to the voting information obtained by each scheduler and the weight of each voting information, Determine the vote score for each scheduler.
  • the arbiter determines a vote score of each scheduler according to the voting information sent by the at least one scheduler, including:
  • the arbiter determines whether the current scheduler votes for other schedulers, the other schedulers are the remaining schedulers in the cloud computing system except the current scheduler; the arbiter determines when the current scheduler votes for other schedulers , the current scheduler's voting score is counted, otherwise, the current scheduler's voting score is determined to be zero.
  • the arbiter determines, according to the voting information sent by the at least one scheduler, a voting score of each scheduler, including: the arbiter directly Count the vote score for each scheduler. In the above manner, the score of each scheduler can be quickly determined.
  • the method further includes: the arbiter receiving the resendment sent by the secondary scheduler
  • the election request the secondary scheduler is a cloud computing system, except for the main scheduler, the re-election request is a secondary scheduler when the current master scheduler does not satisfy the preset condition for the calculation of the computing resources in the cloud computing system.
  • the arbitrator sends a re-election instruction to all the schedulers when the proportion of the secondary scheduler that sends the re-election request satisfies the preset condition, so that all the schedulers re-vomit according to the load status of the current cloud computing system.
  • the main scheduler can be reselected when the current load condition of the cloud computing system does not match the master scheduler, so that the selected master scheduler always matches the load status of the cloud computing system.
  • a method for determining a master scheduler from a cloud computing system includes at least an arbiter and a plurality of schedulers, each scheduler corresponding to a scheduling policy, and the method includes: the scheduler determines the cloud computing The current load status of the system; the scheduler determines the voting information according to the current load status of the cloud computing system and the corresponding relationship between the preset load status and the voting information, wherein, under the current load condition of the cloud computing system, the plurality of schedulers The scheduler corresponding to the voting result determined by the voting information is a scheduler matching the current load status; the scheduler sends The voting information is sent to the arbiter, so that the arbiter determines the primary scheduler according to the voting information, and the primary scheduler is configured to schedule the computing resources in the cloud computing system according to the corresponding scheduling policy.
  • each scheduler can select a scheduler that matches the current load condition, and report the vote information for the matched scheduler, so that the
  • the method further includes: the scheduler periodically collects a load condition of the cloud computing system; and the scheduler does not meet the preset condition when the load condition of the current cloud computing system does not meet the preset condition. And re-voting the voting information to the arbiter, so that the arbiter triggers all the schedulers in the cloud computing system to resend the voting information according to the re-voting information.
  • the main scheduler can be reselected when the current load condition of the cloud computing system does not match the master scheduler, so that the selected master scheduler always matches the load status of the cloud computing system.
  • a third aspect provides an apparatus for determining a master scheduler from a cloud computing system, where the cloud computing system includes at least a plurality of schedulers, each scheduler corresponding to a scheduling policy, and the apparatus includes: a receiving unit, configured to receive at least one The voting information sent by the scheduler, the voting information is generated by the scheduler according to the current load status of the cloud computing system and the corresponding relationship between the preset load status and the voting information; the voting information is that the current scheduler is the multiple scheduling The voting information of the at least one scheduler in the device, the voting information may be specifically an identifier of a voting scheduler of the current scheduler, or the current scheduler is a rating of the voting scheduler; and the selecting unit is configured to send according to the at least one scheduler The voting information is determined from the plurality of schedulers, so that the primary scheduler schedules the computing resources in the cloud computing system according to the corresponding scheduling policy, where the primary scheduler corresponds to the The scheduling policy matches the current load status of the cloud computing system.
  • the selecting unit is configured to: determine a voting score of each scheduler according to the voting information sent by the at least one scheduler; from multiple schedulers
  • the scheduler that determines that the voting score satisfies the preset condition is the master scheduler.
  • the scheduler that meets the preset condition may be selected as the main scheduler, and the master scheduler may schedule the entire cloud computing system according to the corresponding scheduling policy, and the user may set different presets. The condition, and then the different schedulers are selected as the master scheduler, so that in the embodiment of the present invention, the user can flexibly select the master scheduler according to requirements.
  • different priorities are set in advance for different schedulers, and the selecting unit is configured to vote information according to the at least one scheduler.
  • the selecting unit is configured to vote information according to the at least one scheduler.
  • the voting score of each scheduler specifically: setting different weights for the vote information of each scheduler according to the priority of each scheduler; according to the vote information obtained by each scheduler and each The weight of the voting information determines the voting score for each scheduler.
  • the selecting unit when determining the voting score of each scheduler according to the voting information sent by the at least one scheduler, For: determining, for a scheduler, whether the current scheduler votes for other schedulers, and other schedulers are the remaining schedulers except the current scheduler in the cloud computing system; when determining that the current scheduler votes for other schedulers, The voting score of the current scheduler is counted, otherwise, the voting score of the current scheduler is determined to be zero. In the embodiment of the present invention, it is avoided that the scheduler with high priority is always elected as the master scheduler.
  • the arbiter determines, according to the voting information sent by the at least one scheduler, a voting score of each scheduler, including: the arbiter directly The vote score of each scheduler is counted; in the above manner, the score of each scheduler can be quickly determined.
  • the device further includes: a reselecting unit, specifically configured to: receive The re-election request sent by the secondary scheduler, the secondary scheduler is a cloud computing system, and other schedulers other than the primary scheduler, the re-election request is a secondary scheduler, and the current master scheduler does not schedule the computing resources in the cloud computing system.
  • a reselecting unit specifically configured to: receive The re-election request sent by the secondary scheduler, the secondary scheduler is a cloud computing system, and other schedulers other than the primary scheduler, the re-election request is a secondary scheduler, and the current master scheduler does not schedule the computing resources in the cloud computing system.
  • the primary scheduler may be reselected when the current load status of the cloud computing system does not match the primary scheduler, such that the selected primary scheduler always matches the load status of the cloud computing system.
  • a fourth aspect provides an apparatus for determining a primary scheduler from a cloud computing system, the cloud computing system including at least an arbiter, the apparatus comprising: a load determining unit, configured to determine a current load status of the cloud computing system; and a voting information determining unit, And determining, according to a current load condition of the cloud computing system and a corresponding relationship between the preset load condition and the voting information, where the voting information corresponding to the voting result determined by the voting information is under the current load condition of the cloud computing system a scheduler matching the current load condition; a sending unit, configured to send the voting information to the arbiter, so that the arbiter determines the primary scheduler according to the voting information, and the primary scheduler is configured to use the corresponding scheduling policy in the cloud computing system
  • the computing resources are scheduled.
  • each scheduler may select a scheduler that matches the current load condition, and report the vote information to the matched scheduler, so that the master scheduler selected by the arbiter and the entire cloud computing system The load conditions match
  • the device further includes: an acquiring unit, configured to periodically collect a load condition of the cloud computing system; and a reporting unit, configured to load the current cloud computing system
  • the preset condition is not met, the re-voting information is reported to the arbiter, so that the arbiter triggers all the schedulers in the cloud computing system to resend the voting information according to the re-voting information.
  • the primary scheduler may be reselected when the current load status of the cloud computing system does not match the primary scheduler, such that the selected primary scheduler always matches the load status of the cloud computing system.
  • a cloud computing system in a fifth aspect, includes an arbiter and a scheduler; wherein the scheduler is configured to determine a current load status of the cloud computing system, and according to a current load status of the cloud computing system, preset Corresponding relationship between the load status and the voting information, determining the voting information and transmitting the voting information to the arbiter; the arbiter is configured to receive the voting information sent by the at least one scheduler, and the plurality of schedulings according to the voting information sent by the at least one scheduler Selecting a scheduler as a master scheduler, so that the master scheduler schedules the computing resources in the cloud computing system according to the corresponding scheduling policy, wherein the scheduling policy corresponding to the master scheduler is The current load conditions of the cloud computing system match.
  • the scheduling policy corresponding to the master scheduler is The current load conditions of the cloud computing system match.
  • the scheduler first determines the current load status of the entire cloud computing system, and then determines the voting information according to the current load status.
  • the arbiter selects a scheduler matching the current load status from the plurality of schedulers of the cloud computing system according to the respective voting information of the plurality of schedulers, and the master scheduler according to the corresponding scheduler
  • the strategy schedules computing resources in the cloud computing system. Therefore, the master scheduler is determined by the arbiter, and the secondary scheduler except the master schedule in the entire cloud computing system will no longer schedule the computing resources, thereby ensuring that the entire cloud computing system only needs the scheduler scheduling.
  • a scheduler is performing scheduling work, thereby avoiding scheduling conflicts between computing resources among different schedulers.
  • the selected scheduler can adapt to the load condition of the current system, thereby improving the targeting of the main scheduler and enhancing the cloud computing system. Scheduling performance.
  • the arbiter since the master scheduler is generated according to the vote of each scheduler, the arbiter only needs to determine the master scheduler according to the vote information, and does not need to select according to the characteristics of each scheduler, thereby realizing the arbiter and each scheduler.
  • the decoupling between the arbiter and the different schedulers is uniformly scheduled, and the scheduler can flexibly change, increase or decrease, and improve the flexibility of the system.
  • FIG. 1 is a schematic diagram of a cloud computing system according to an embodiment of the present invention.
  • FIG. 2 is another schematic diagram of a cloud computing system according to an embodiment of the present invention.
  • FIG. 3 is a schematic diagram of a cloud computing system according to an embodiment of the present invention.
  • FIG. 4 is a schematic diagram of a method for determining a master scheduler in a cloud computing system according to an embodiment of the present invention
  • FIG. 5 is a schematic diagram of a cloud computing system according to an embodiment of the present disclosure.
  • FIG. 6 is a schematic diagram of a cloud computing system according to an embodiment of the present disclosure.
  • FIG. 7 is a schematic diagram of a cloud computing system according to an embodiment of the present disclosure.
  • FIG. 8 is a schematic diagram of an apparatus for determining a master scheduler in a cloud computing system according to an embodiment of the present disclosure
  • FIG. 9 is a schematic diagram of an apparatus for determining a master scheduler in a cloud computing system according to an embodiment of the present invention.
  • FIG. 10 is a schematic diagram of a cloud computing system according to an embodiment of the present disclosure.
  • FIG. 11 is a schematic diagram of an arbitrator according to an embodiment of the present invention.
  • FIG. 12 is a schematic diagram of a scheduler according to an embodiment of the present invention.
  • the present invention provides a method and apparatus for determining a master scheduler from a cloud computing system. As shown in FIG. 1 , the application scenarios of the method and apparatus are as follows:
  • cloud computing is a computing model.
  • the whole cloud computing system is mainly composed of a client, a scheduler and a resource pool.
  • the scheduler can receive The request of the client, and according to the request of the client, schedules the computing resources in the resource pool for use by the client.
  • the cloud computing system shown in FIG. 1 it is usually a single scheduler, that is, there is only one scheduler in a cloud computing system; but as the types of services carried by the cloud computing system are more and more, in the cloud computing system
  • the scheduler also gradually evolves from a single scheduler to a multi-scheduler; wherein, in a multi-scheduler cloud computing system, each scheduler corresponds to a scheduling policy for scheduling computing resources in the cloud computing system.
  • different schedulers work independently, that is, different schedulers independently schedule computing resources in a cloud computing system, and therefore, using the above method, It may cause conflicts between computing resources scheduled by different schedulers.
  • the present invention provides a method for determining a master scheduler from a cloud computing system, wherein the core idea of the method is: first determining a current load status of the entire cloud computing system, and then, from the cloud computing system Among the multiple schedulers, the scheduler matching the current load condition is selected as the master scheduler, and the master scheduler will schedule the computing resources in the cloud computing system according to the corresponding scheduling policy, and the entire cloud computing
  • the secondary scheduler in the system except the primary scheduler will no longer schedule the computing resources, so that at each moment, only one scheduler of the entire cloud computing system is performing scheduling, thereby avoiding scheduling resources between different schedulers.
  • conflict issue is: first determining a current load status of the entire cloud computing system, and then, from the cloud computing system Among the multiple schedulers, the scheduler matching the current load condition is selected as the master scheduler, and the master scheduler will schedule the computing resources in the cloud computing system according to the corresponding scheduling policy, and the entire cloud computing
  • the present invention first provides a cloud computing system.
  • the system includes at least an arbiter, a plurality of schedulers, and a resource pool:
  • the resource pool is used to provide computing resources; each scheduler corresponds to a scheduling policy, which is used to schedule computing resources in the resource pool according to its corresponding scheduling policy; the arbiter is used to select from multiple schedulers.
  • a scheduler as a master scheduler; wherein the scheduler selected as the master scheduler will schedule the computing resources in the resource pool according to its corresponding scheduling policy, and the scheduler not selected as the master scheduler is called
  • the secondary scheduler stops scheduling the computing resources in the resource pool; for example, the entire cloud computing system includes three schedulers, namely, scheduler A, scheduler B, and scheduler C, wherein scheduler A corresponds to a scheduling policy.
  • scheduler B corresponds to scheduling policy B
  • scheduler C corresponds to scheduling policy C
  • the arbiter selects scheduler A as the master scheduler
  • scheduler B and scheduler C will be the secondary scheduler
  • scheduler A will schedule computing resources in the resource pool according to its corresponding scheduling policy A
  • scheduler B and scheduler C will act as secondary schedulers, stopping resources. The computing resources in the pool are scheduled.
  • the scheduling method provided by the present invention is specifically as follows:
  • Step S41 The scheduler acquires a current load status of the resource pool in the entire cloud computing system.
  • the current load status of the resource pool refers to the load status of various resources of each server in the current resource pool.
  • the current load status can be reflected as some specific parameters. When the parameter meets certain conditions (for example, a specific threshold is met), the resource is considered as a resource.
  • the load status of the pool is in a specific load condition; it is also possible to make the resource pool meet the conditions of a certain load condition by monitoring whether a specific trigger event occurs in the resource pool.
  • the load status may refer to the total load status of resources of all servers in the resource pool, for example, the total amount of load of all servers in the resource pool of one or more resources, or the total amount of resources.
  • the load condition may also determine whether a certain load condition is preset by whether a part of the server triggers a specific condition, for example, the resource load of some servers is greater than a specific threshold (available to Judging the balanced load of the servers in the resource pool), or the certain types of resources of some servers exceed a certain threshold (which can be used to determine the balance of resource types allocated by servers in the resource pool).
  • the resource pool load status can be collected actively by the scheduler, or collected by the functional nodes in other cloud computing systems and sent to the scheduler.
  • the collection or detection of the resource load status can be determined by the prior art, for example, according to the resource load information actively reported by each server, or determined by the management node when the task is allocated, or by monitoring the data traffic between the servers. Wait.
  • Step S42 The scheduler determines the voting information according to the current load status of the cloud computing system and the corresponding relationship between the preset load status and the voting information.
  • the correspondence between the preset load status and the voting information in each scheduler may be the same or different, and the related technical personnel may set the correspondence between the load status and the voting information according to the requirements.
  • the voting information is that the current scheduler is voting information of at least one of the plurality of schedulers, and the voting information may be specifically an identifier of the voted scheduler, or is voted and scheduled.
  • the corresponding relationship between the preset load status and the voting information is “Load 1—Scheduler 1, Load 2—Scheduler 2, and Load 3—Scheduler 3”,
  • the scheduler matched by the current scheduler is "scheduler 1”
  • the "scheduler 1" is specifically the identifier of the voted scheduler, and, for example,
  • the corresponding relationship between the load status and the voting information is "load 1 - scheduler 1 rating 5 points, load 2 - scheduler 2 rating 6 points, load 3 - scheduler 3 rating 10 points", then in the entire cloud computing system
  • the voting information matched by the scheduler is "Scheduler 1 rating
  • Step S43 The scheduler reports the voting information to the arbitrator
  • Step S44 The arbiter determines the master scheduler from the plurality of schedulers according to the vote information of each scheduler, and the remaining schedulers serve as the secondary scheduler, wherein the master scheduler will follow the corresponding scheduling policy in the resource pool.
  • the computing resource is scheduled, and the secondary scheduler stops scheduling the computing resources in the resource pool.
  • the following method may be specifically used to select a primary scheduler from multiple schedulers:
  • the arbiter determines a vote score of each scheduler according to the vote information of each scheduler, and then, from among the plurality of schedulers, determines a scheduler whose vote score satisfies a preset condition as a master scheduler, and
  • the preset condition that the scheduler score satisfies can be set by a person skilled in the art, for example, the preset condition can be set to use the scheduler with the highest score as the master scheduler, or the preset condition can be set as the scheduler with the lowest score.
  • the voting condition may be set as the master scheduler, and the scheduler that scores the voting score in a certain interval is not described here.
  • the process of selecting the main scheduler is described in detail by taking the “predetermined condition as the scheduler with the highest score as the main scheduler” as an example:
  • the method may specifically correspond to the voting information being the identifier of the voted scheduler, specifically: manually setting different priorities for each scheduler, and then the arbiter receives each scheduler After voting information, different weights are set for each voting information according to the priority of the scheduler that sends the voting information; then, according to the voting information obtained by each scheduler and the weight of each voting information, each scheduler is determined. Voting score; finally, from among the plurality of schedulers, the scheduler with the highest voting score is selected as the main scheduler;
  • the entire cloud computing system includes three schedulers of schedulers A, B, and C, and the priority of scheduler A is set to high in advance, and the priority of scheduler B is set to medium, and scheduling is performed.
  • the method may correspond to the voting information being the rating of the voting scheduler.
  • the arbiter directly counts the score of each scheduler.
  • the above example is still used, and the entire cloud computing system, including the scheduler.
  • the voter A's voting information is "Scheduler B rating 5 points”
  • the scheduler B's voting information is "Scheduler C rating 3 points”
  • “Scheduler C vote The information is "Scheduler C rating 6 points”
  • the statistic scheduler B's voting score is 5 points
  • the scheduler A's voting score is 0. Minute.
  • this method can correspond to the voting information being the identifier of the voted scheduler and the rating of the voted scheduler.
  • each scheduler sends at least two voting information to the arbiter, one of which votes.
  • the information is itself, indicating that the current scheduler is to participate in the election, and the other vote is the other scheduler, and the other schedulers are the remaining schedulers in the cloud computing system except the current scheduler;
  • the arbiter After receiving the voting information of the scheduler, the arbiter first determines whether the current scheduler votes for other schedulers. If voting for other schedulers, the current scheduler's voting score is counted. Otherwise, the current scheduler's score is determined to be zero. ;
  • the method disclosed in the first manner above may be specifically adopted, and the second common addition method may also be adopted;
  • the arbiter judges the voter's voting score to be zero, thereby avoiding a situation in which a scheduler only votes for itself.
  • the current load status of the entire cloud computing system is first determined, and then, from the plurality of schedulers of the cloud computing system, a scheduler matching the current load status is selected as the primary scheduler.
  • the master scheduler will schedule the computing resources in the cloud computing system according to the corresponding scheduling policy, and the secondary scheduler in the entire cloud computing system except the primary scheduling will no longer schedule the computing resources, thereby ensuring At each moment, only one scheduler of the entire cloud computing system is performing scheduling work, thereby avoiding scheduling conflicts between computing resources among different schedulers.
  • the primary scheduler after the primary scheduler is selected in the entire cloud computing system by using the method disclosed in Embodiment 1, the primary scheduler will schedule the computing resources in the resource pool according to the corresponding scheduling policy.
  • the secondary scheduler that is not selected as the master scheduler will monitor the load status of the entire cloud computing system in real time, and report the re-voting information to the arbitrator when the load condition does not satisfy the preset condition; and the arbitrator sends a re-voting
  • the re-election instruction is sent to all the schedulers, and the scheduler will resend the voting information in the manner disclosed in Embodiment 1, and the arbiter will re-elect the main scheduler.
  • the elected master scheduler matches the current load status of the cloud computing system.
  • the resource pool may be composed of multiple servers, each of which provides a certain computing resource, and the foregoing three schedulers may be specifically LB (Load Balanc) scheduling.
  • the LC (Load Consolidation) scheduler and the HE (Hotspot Elimination) scheduler wherein the scheduling policy corresponding to the LB scheduler is an LB scheduling policy, and the scheduling policy corresponding to the LC scheduler is an LC scheduling
  • the policy, the scheduling policy corresponding to the HE scheduler is the HE scheduling policy;
  • the foregoing LB scheduling policy may specifically ensure that each server in the resource pool is load balanced when performing service allocation to the server;
  • the LC scheduling policy may specifically: when the load of a server in the resource pool is lower than a preset minimum value, the service of the server is migrated to other servers in the resource pool, and the server is powered off;
  • the HE scheduling policy may specifically be when the load of a server in the resource pool exceeds a preset maximum value. Part of the business is migrated to other servers in the resource pool, reducing the load on the current server.
  • the priority of the HE scheduler is set to high in advance
  • the priority of the LB scheduler is set to medium
  • the priority of the LC scheduler is set to low
  • the high priority HE scheduler is set.
  • the weight of the voting information is set to 3
  • the weight of the voting information of the medium priority LB scheduler is set to 2
  • the weight of the voting information of the low priority LC scheduler is set to 1;
  • a voting policy inside each scheduler can be set by a person skilled in the art. It is assumed that, in the embodiment of the present invention, the voting strategies of the HE, LB, and LC schedulers are as follows:
  • Voting strategy of the HE scheduler when the current load of the cloud computing system is greater than Th, it votes for HE;
  • the current load of the cloud computing system is greater than or equal to T1, and when it is less than or equal to Th, it is voted for LB;
  • Voting strategy of the LB scheduler when the current load of the cloud computing system is less than T1, it votes for the LC;
  • the voting strategy of the LC scheduler when the current load of the cloud computing system is less than T1 and lasts for m minutes, it votes for the LC;
  • the voting information of the HE scheduler may be determined as “voting for LC”, and the LB scheduler The voting information is “voting for LC”, and the voting information of the LC scheduler is “voting for LC”;
  • the LC and LB schedulers will monitor the current load status of the cloud computing system in real time, and when the current load of the cloud computing system does not satisfy the condition, for example (when the current load is greater than Th)
  • the LB and LC scheduler may send a re-election request to the arbiter, and the arbiter triggers all schedulers to re-elect when the proportion of the re-election scheduler is greater than a certain value; for example, when sending a re-election scheduler When the proportion accounts for 60% of all schedulers in the cloud system, the election is re-elected.
  • the entire cloud computing system includes three schedulers, an arbiter, and a resource pool, and the three schedulers are a scientific computing scheduler, a storage cloud scheduler, and a web service.
  • the scheduler is taken as an example to describe in detail the process of the present invention:
  • the scientific computing scheduler is used for scientific computing services, and the computing resources are scheduled.
  • the corresponding scheduling strategy is to provide sufficient CPU resources for the scientific computing service, and the corresponding priority is 5;
  • the storage cloud scheduler is used to store the cloud service.
  • Scheduling computing resources, the corresponding scheduling strategy is to balance disk input and output and disk capacity as much as possible, and reduce fragmentation, and its corresponding priority is 3;
  • the Web service scheduler is used for Web services, scheduling computing resources, and corresponding tuning
  • the degree policy is to provide sufficient CPU and network resources for the current service, and the corresponding priority is 7;
  • the three loaders may collect the current load status of the cloud computing system, and the current load of the cloud computing system may specifically include the disk load of the cloud computing system and the network load;
  • the scheduling policies of the three schedulers may be as follows:
  • Scientific computing scheduler first, vote for itself; second, vote for the storage cloud scheduler when the remaining disk resources in the resource pool are higher than Rs, otherwise vote for the Web service scheduler;
  • Storage cloud scheduler First, vote for itself; second, vote for the Web service scheduler when the remaining network resources in the resource pool are lower than Rw, otherwise vote for the scientific computing scheduler.
  • Web service scheduler first, vote for itself; second, vote for the storage cloud scheduler when the remaining network resources in the resource pool are higher than Rs, otherwise vote for the scientific calculation scheduler.
  • each scheduler can vote for two votes, one vote for itself, indicating that the current scheduler participates in the election, and another ticket is voted for other schedules that match the current load status. Device.
  • the remaining disk resources in the cloud computing system are higher than Rs, and the remaining network resources are lower than Rw as an example, and the process of the present invention is described in detail:
  • the voting information for the scientific computing scheduler may be specifically: voting for itself, that is, the scientific computing scheduler, voting for the storage cloud scheduler; and for the storage cloud scheduler
  • the voting information may be specifically: voting for itself, that is, voting for the storage cloud scheduler, voting for the Web service scheduler; voting information for the Web service scheduler may be specifically: voting for itself, that is, voting for the Web service scheduler, Vote for the storage cloud scheduler.
  • the arbiter may first determine whether the current scheduler votes for itself, and if it votes for itself, determine that the current scheduler has the qualification for the election; otherwise, determine the current scheduler. No eligibility, no longer count the score of the current scheduler; then, determine whether each scheduler votes for other schedulers, if the current scheduler votes for other schedulers, count the score of the current scheduler, otherwise, determine the current The score of the scheduler is zero. In the above manner, the scheduler with high priority can be avoided to vote only for itself, but not for other schedulers, so that the scheduler with high priority is always elected as the master scheduler.
  • the election storage cloud scheduler is the main scheduler, and the scientific calculation scheduler and the Web service scheduler are scheduled.
  • the device is a secondary scheduler; and the storage cloud scheduler will schedule the computing resources in the resource pool according to its corresponding scheduling policy, and the scientific computing scheduler and the Web service scheduler as the secondary scheduler can monitor the current resource pool in real time.
  • the load status including the disk resource status and the network resource status; and each scheduler sends a re-election request to the arbiter upon discovering that the current load status of the resource pool does not satisfy its own condition, and the arbiter
  • the scheduler that sends the re-election request reaches a certain proportion, all schedulers are triggered to re-elect.
  • the present invention also provides an apparatus for determining a master scheduler from a cloud computing system. As shown in FIG. 8, the apparatus includes:
  • the receiving unit 81 is configured to receive voting information sent by at least one scheduler, where the voting information is generated by the scheduler according to a current load status of the cloud computing system and a corresponding relationship between a preset load status and voting information. ;
  • the selecting unit 82 is configured to determine, according to the voting information sent by the at least one scheduler, a master scheduler from the plurality of schedulers, so that the master scheduler pairs the cloud computing system according to the corresponding scheduling policy.
  • the computing resources in the scheduling are performed, wherein the scheduling policy corresponding to the primary scheduler matches the current load status of the cloud computing system.
  • the selecting unit is specifically configured to determine, according to the voting information sent by the at least one scheduler, a voting score of each scheduler; and from the plurality of schedulers, determine a scheduler that the voting score satisfies a preset condition For the main scheduler.
  • the selecting unit is configured to: when determining a voting score of each scheduler according to the voting information sent by the at least one scheduler, specifically:
  • each scheduler different weights are set for the voting information of each scheduler; the voting score of each scheduler is determined according to the voting information obtained by each scheduler and the weight of each voting information.
  • the selecting unit when determining the voting score of each scheduler according to the voting information sent by the at least one scheduler, is specifically used to: determine, for a scheduler, whether the current scheduler is another scheduler Voting, the other scheduler is the remaining scheduler except the current scheduler in the cloud computing system; when determining that the current scheduler votes for other schedulers, the current scheduler's voting score is counted; otherwise, the current schedule is determined.
  • the voter's vote score is zero.
  • the device further includes: a reselecting unit, configured to: receive a re-election request sent by the secondary scheduler, where the secondary scheduler is in the cloud computing system, except the main scheduler a scheduler, where the re-election request is sent when the current master scheduler does not satisfy the preset condition for the calculation of the computing resource in the cloud computing system; in the secondary scheduler that sends the re-election request When the ratio meets the preset condition, a re-election instruction is sent to all schedulers, so that all schedulers re-vomit according to the load status of the current cloud computing system.
  • a reselecting unit configured to: receive a re-election request sent by the secondary scheduler, where the secondary scheduler is in the cloud computing system, except the main scheduler a scheduler, where the re-election request is sent when the current master scheduler does not satisfy the preset condition for the calculation of the computing resource in the cloud computing system; in the secondary scheduler that sends the re
  • the current load status of the entire cloud computing system is first determined, and then, from the plurality of schedulers of the cloud computing system, a scheduler matching the current load status is selected as the primary scheduler.
  • the master scheduler will schedule the computing resources in the cloud computing system according to the corresponding scheduling policy, and the secondary scheduler in the entire cloud computing system except the primary scheduling will no longer schedule the computing resources, thereby ensuring At each moment, only one scheduler of the entire cloud computing system is performing scheduling work, thereby avoiding scheduling conflicts between computing resources among different schedulers.
  • the present invention also provides an apparatus for determining a master scheduler from a cloud computing system. As shown in FIG. 9, the apparatus includes:
  • a load determining unit 91 configured to determine a current load status of the cloud computing system
  • a voting information determining unit 92 configured to determine a current load status of the cloud computing system and a preset load status Corresponding to the voting information, determining the voting information, wherein, under the current load condition of the cloud computing system, the scheduler corresponding to the voting result determined by the voting information is a scheduler matching the current load status;
  • a sending unit 93 configured to send the voting information to an arbiter, so that the arbiter determines, according to the voting information, a primary scheduler, where the primary scheduler is configured to be in the cloud computing system according to a corresponding scheduling policy.
  • the computing resources are scheduled.
  • the device further includes: an collecting unit, configured to periodically collect a load condition of the cloud computing system; and a reporting unit, configured to report when the current cloud computing system does not meet the preset condition
  • the current load status of the entire cloud computing system is first determined, and then, from the plurality of schedulers of the cloud computing system, a scheduler matching the current load status is selected as the primary scheduler.
  • the master scheduler will schedule the computing resources in the cloud computing system according to the corresponding scheduling policy, and the secondary scheduler in the entire cloud computing system except the primary scheduling will no longer schedule the computing resources, thereby ensuring At each moment, only one scheduler of the entire cloud computing system is performing scheduling work, thereby avoiding scheduling conflicts between computing resources among different schedulers.
  • the present invention further provides a cloud computing system, as shown in FIG. 10, the cloud computing system includes an arbiter 101 and a plurality of schedulers 102;
  • the scheduler 102 is configured to determine a current load status of the cloud computing system, and determine a voting information according to a current load status of the cloud computing system, a corresponding relationship between a preset load status and voting information, and send the Voting information to the arbiter;
  • the arbiter 101 is configured to receive voting information sent by at least one scheduler, and select a scheduler from the plurality of schedulers as a primary scheduler according to the voting information sent by the at least one scheduler, so that the primary scheduling
  • the computing resource in the cloud computing system is scheduled according to the corresponding scheduling policy, wherein the scheduling policy corresponding to the primary scheduler matches the current load status of the cloud computing system.
  • the current load status of the entire cloud computing system is first determined, and then, from the plurality of schedulers of the cloud computing system, a scheduler matching the current load status is selected as the primary scheduler.
  • the master scheduler will schedule the computing resources in the cloud computing system according to the corresponding scheduling policy, and the secondary scheduler in the entire cloud computing system except the primary scheduling will no longer schedule the computing resources, thereby ensuring At each moment, only one scheduler of the entire cloud computing system is performing scheduling work, thereby avoiding scheduling conflicts between computing resources among different schedulers.
  • the present invention also provides an arbiter, as shown in Figure 11, the arbiter includes at least a memory 111 and a processor 112;
  • a memory 111 for storing programs and instructions
  • the processor 112 is configured to execute: by calling a program and an instruction stored in the memory:
  • the voting information is generated by the scheduler according to a current load status of the cloud computing system and a corresponding relationship between the preset load status and the voting information;
  • the present invention further provides a scheduler, the scheduler including at least a memory 121 and a processor 122;
  • a memory 121 for storing programs and instructions
  • the processor 122 is configured to execute by calling a program and an instruction stored in the memory:
  • the determined scheduler is a scheduler that matches the current load condition
  • the bus architecture may include any number of interconnected buses and bridges, specifically linked by one or more processors represented by the processor and various circuits of memory represented by the memory.
  • the bus architecture can also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and, therefore, will not be further described herein.
  • the bus interface provides an interface.
  • the processor is responsible for managing the bus architecture and the usual processing, and the memory can store the data that the processor uses when performing operations.
  • the current load status of the entire cloud computing system is first determined, and then, from the plurality of schedulers of the cloud computing system, a scheduler matching the current load status is selected as the primary scheduler.
  • the master scheduler will schedule the computing resources in the cloud computing system according to the corresponding scheduling policy, and the secondary scheduler in the entire cloud computing system except the primary scheduling will no longer schedule the computing resources, thereby ensuring At each moment, only one scheduler of the entire cloud computing system is performing scheduling work, thereby avoiding scheduling conflicts between computing resources among different schedulers.
  • embodiments of the present invention can be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or a combination of software and hardware. Moreover, the invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.
  • computer-usable storage media including but not limited to disk storage, CD-ROM, optical storage, etc.
  • the computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device.
  • the apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.
  • These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device.
  • the instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.

Abstract

本申请公开了一种从云计算系统中确定主调度器的方法及装置，所述云计算系统至少包括仲裁器和多个调度器，每个调度器对应一种调度策略，其中，所述方法包括：所述仲裁器接收至少一个调度器发送的投票信息，所述投票信息为所述调度器根据所述云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系所生成的；所述仲裁器根据所述至少一个调度器发送的投票信息，从所述多个调度器中确定主调度器，以使得所述主调度器按照所对应的调度策略对所述云计算系统中的计算资源进行调度，其中，所述主调度器所对应的所述调度策略与所述云计算系统的当前负载状况相匹配；采用本发明的方法及装置，可避免不同调度器间所分配计算资源的冲突。

Description

从云计算系统中确定主调度器的方法及装置
本申请要求在2016年11月30日提交中国专利局、申请号为201611089191.9、发明名称为“从云计算系统中确定主调度器的方法及装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本发明涉及云计算技术领域，尤其涉及一种从云计算系统中确定主调度器的方法及装置。
背景技术
随着互联网的飞速发展，云计算应运而生；所谓云计算是一种计算模式，如图1所示，整个云计算系统主要由客户端、调度器和资源池组成；其中，调度器可接收客户端的请求，且根据客户端的请求，调度资源池中的计算资源给客户端使用。
目前，在图1所示的云计算系统中通常为单调度器，也就是一个云计算系统中只有一个调度器；但是随着云计算系统所承载业务的类型越来越多，云计算系统中的调度器也逐渐由单调度器演变为多调度器；其中，在多调度器的云计算系统中，每个调度器对应一种调度策略，负责一种类型的业务；比如，一云计算系统承载三种类型的业务，分别为搜索引擎、视频监控以及大数据；那么，如图2所示，整个云计算系统中，可有三个调度器，分别为搜索引擎调度器、视频监控调度器以及大数据调度器；相应的，搜索引擎调度器对应搜索引擎的调度策略，负责管理搜索引擎客户端的业务，即搜索引擎调度器接收搜索引擎客户端的业务请求，且调度资源池中的计算资源给搜索引擎客户端；同理，视频监控调度器对应视频监控调度策略，负责视频监控客户端的业务，大数据调度器对应大数据的调度策略，负责大数据客户端的业务，在此不再赘述。
由于在实际应用中，在多调度器的云计算系统中，不同调度器间是各自独立的，且不同调度器是共用资源池中的计算资源的，因此采用上述方法，可能会造成不同调度器所分配计算资源的冲突，仍沿用图2所示的举例，比如，搜索引擎调度器在接收到搜索引擎客户端的业务请求时，分配资源池中的计算资源M给搜索引擎客户端；而由于搜索引擎调度器与大数据调度器彼此间是相互独立的，可能会出现，在大数据服务器接收到大数据客户端的业务请求时，也分配资源池中的计算资源M给大数据客户端，从而会造成所分配计算资源的冲突。
发明内容
本发明实施例提供一种从云计算系统中确定主调度器的方法及装置，以避免不同调度器间所分配计算资源的冲突。
第一方面，提供一种从云计算系统中确定主调度器的方法，云计算系统至少包括仲裁器和多个调度器，每个调度器对应一种调度策略，方法包括：仲裁器接收至少一个调度器发送的投票信息，投票信息为调度器根据云计算系统的当前负载状况以及预设的负 载状况与投票信息的对应关系所生成的；所述投票信息为当前调度器为所述多个调度器中至少一个调度器的投票信息，所述投票信息可具体为被投票调度器的标识，或，被投票调度器的评分；仲裁器根据至少一个调度器发送的投票信息，从所述多个调度器中确定主调度器，以使得主调度器按照所对应的调度策略对云计算系统中的计算资源进行调度，其中，主调度器所对应的所述调度策略与云计算系统的当前负载状况相匹配。
结合第一方面，在第一种可能的实现方式中，仲裁器根据至少一个调度器发送的投票信息，从多个调度器中选择一调度器作为主调度器，包括：仲裁器根据至少一个调度器发送的投票信息，确定每个调度器的投票得分；仲裁器从多个调度器中，确定投票得分满足预设条件的调度器为所述主调度器。由于在本发明实施例中，可选择满足预设条件的调度器为主调度器，而主调度器可根据其对应的调度策略对整个云计算系统进行调度，那么用户可通过设置不同的预设条件，进而选择不同的调度器作为主调度器，从而在本发明实施例中，可实现用户根据需求灵活地选择主调度器。
结合第一方面的第一种可能实现方式，在第二种可能的实现方式中，预先为不同调度器设置不同的优先级，仲裁器根据至少一个调度器发送的投票信息，确定每个调度器的投票得分，包括：仲裁器根据每个调度器的优先级，为每个调度器的投票信息设置不同的权重；仲裁器根据每个调度器所获得的投票信息以及每个投票信息的权重，确定每个调度器的投票得分。采用上述方法，可划分不同调度器的优先级，从而使得优先级高的调度器所占投票得分的比重较大，进而使得统计投票得分的方式更加实用。
结合第一方面的第一种可能实现方式，在第三种可能的实现方式中，仲裁器根据至少一个调度器发送的投票信息，确定每个调度器的投票得分，包括：
针对一调度器，仲裁器确定当前调度器是否为其它调度器投票，其它调度器为云计算系统中，除当前调度器外的剩余调度器；仲裁器在确定当前调度器为其它调度器投票时，统计当前调度器的投票得分，否则，确定当前调度器的投票得分为零。采用上述方法，可避免优先级高的调度器总是被当选为主调度器。
结合第一方面的第一种可能实现方式，在第四种可能的实现方式中，仲裁器根据至少一个调度器发送的投票信息，确定每个调度器的投票得分，包括：所述仲裁器直接统计每个调度器的投票得分。采用上述方式，可快速地确定每个调度器的得分。
结合第一方面，第一方面的第一种可能实现方式、第二种可能实现方式、第三种可能实现方式或第四种可能实现方式，方法还包括：仲裁器接收辅调度器发送的重新选举请求，辅调度器为云计算系统中，除主调度器外的其它调度器，重新选举请求为辅调度器在当前主调度器对云计算系统中计算资源的调度不满足预设条件时所发送的；仲裁器在发送重新选举请求的辅调度器的比例满足预设条件时，向所有调度器发送重新选举指令，以使得所有调度器根据当前云计算系统的负载状况，重新进行投票。采用上述方法，可在云计算系统的当前负载状况与主调度器不匹配时，重新选择主调度器，从而使得所选择的主调度器总是与云计算系统的负载状况相匹配。
第二方面，提供一种从云计算系统中确定主调度器的方法，云计算系统至少包括仲裁器和多个调度器，每个调度器对应一种调度策略，方法包括：调度器确定云计算系统的当前负载状况；调度器根据云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系，确定投票信息，其中，在云计算系统的当前负载状况下，多个调度器的投票信息所确定的投票结果对应的调度器为与当前负载状况所匹配的调度器；调度器发 送投票信息至仲裁器，以使得仲裁器根据投票信息，确定主调度器，主调度器用于按照对应的调度策略对云计算系统中的计算资源进行调度。采用上述方法，每个调度器可选择与当前负载状况相匹配的调度器，并为相匹配的调度器上报投票信息，从而使得仲裁器所选择的主调度器与整个云计算系统的负载状况相匹配。
结合第二方面，在第二方面的第一种可能实现方式中，方法还包括：调度器周期性采集云计算系统的负载状况；调度器在当前云计算系统的负载状况不满足预设条件时，上报重新投票信息至仲裁器，以使得仲裁器根据重新投票信息，触发云计算系统中的所有调度器重新发送投票信息。采用上述方法，可在云计算系统的当前负载状况与主调度器不匹配时，重新选择主调度器，从而使得所选择的主调度器总是与云计算系统的负载状况相匹配。
第三方面，提供一种从云计算系统中确定主调度器的装置，云计算系统至少包括多个调度器，每个调度器对应一种调度策略，装置包括：接收单元，用于接收至少一个调度器发送的投票信息，投票信息为调度器根据云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系所生成的；所述投票信息为当前调度器为所述多个调度器中至少一个调度器的投票信息，所述投票信息可具体为当前调度器所投票调度器的标识，或，当前调度器为投票调度器的评分；选择单元，用于根据至少一个调度器发送的投票信息，从所述多个调度器中确定主调度器，以使得主调度器按照所对应的调度策略对云计算系统中的计算资源进行调度，其中，所述主调度器所对应的所述调度策略与所述云计算系统的当前负载状况相匹配。结合第三方面，在第三方面的第一种可能实现方式中，选择单元，具体用于：根据至少一个调度器发送的投票信息，确定每个调度器的投票得分；从多个调度器中，确定投票得分满足预设条件的调度器为所述主调度器。由于在本发明实施例中，可选择满足预设条件的调度器为主调度器，而主调度器可根据其对应的调度策略对整个云计算系统进行调度，那么用户可通过设置不同的预设条件，进而选择不同的调度器作为主调度器，从而在本发明实施例中，可实现用户根据需求，灵活的选择主调度器。
结合第三方面的第一种可能实现方式中，在第三方面的第二种可能实现方式中，预先为不同调度器设置不同的优先级，选择单元，在根据至少一个调度器发送的投票信息，确定每个调度器的投票得分时，具体用于：根据每个调度器的优先级，为每个调度器的投票信息设置不同的权重；根据每个调度器所获得的投票信息以及每个投票信息的权重，确定每个调度器的投票得分。采用上述方法，可划分不同调度器的优先级，从而使得优先级高的调度器所占投票得分的比重较大，进而使得统计投票得分的方式更加实用。第三方面的第一种可能实现方式中，在第三方面的第三种可能实现方式中，选择单元，在根据至少一个调度器发送的投票信息，确定每个调度器的投票得分时，具体用于：针对一调度器，确定当前调度器是否为其它调度器投票，其它调度器为云计算系统中，除当前调度器外的剩余调度器；在确定当前调度器为其它调度器投票时，统计当前调度器的投票得分，否则，确定当前调度器的投票得分为零。在本发明实施例中，可避免优先级高的调度器总是被当选为主调度器。
结合第三方面的第一种可能实现方式，在第四种可能的实现方式中，仲裁器根据至少一个调度器发送的投票信息，确定每个调度器的投票得分，包括：所述仲裁器直接统计每个调度器的投票得分；采用上述方式，可快速的确定每个调度器的得分。
结合第三方面，第三方面的第一种可能实现方式、第二种可能实现方式、第三种可能实现方式或第四种可能实现方式，装置还包括，重新选择单元，具体用于：接收辅调度器发送的重新选举请求，辅调度器为云计算系统中，除主调度器外的其它调度器，重新选举请求为辅调度器在当前主调度器对云计算系统中计算资源的调度不满足预设条件时所发送的；在发送重新选举请求的辅调度器的比例满足预设条件时，向所有调度器发送重新选举指令，以使得所有调度器根据当前云计算系统的负载状况，重新进行投票。在本发明实施例中，可在云计算系统的当前负载状况与主调度器不匹配时，重新选择主调度器，从而使得所选择的主调度器总是与云计算系统的负载状况相匹配。
第四方面，提供一种从云计算系统中确定主调度器的装置，云计算系统至少包括仲裁器，装置包括：负载确定单元，用于确定云计算系统的当前负载状况；投票信息确定单元，用于根据云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系，确定投票信息，其中，在云计算系统的当前负载状况下，投票信息所确定的投票结果对应的调度器为与当前负载状况所匹配的调度器；发送单元，用于发送投票信息至仲裁器，以使得仲裁器根据投票信息，确定主调度器，主调度器用于按照对应的调度策略对云计算系统中的计算资源进行调度。在本发明实施例中，每个调度器可选择与当前负载状况相匹配的调度器，并为相匹配的调度器上报投票信息，从而使得仲裁器所选择的主调度器与整个云计算系统的负载状况相匹配。
结合第四方面，在第四方面的第一种可能实现方式中，装置还包括：采集单元，用于周期性采集云计算系统的负载状况；上报单元，用于在当前云计算系统的负载状况不满足预设条件时，上报重新投票信息至仲裁器，以使得仲裁器根据重新投票信息，触发云计算系统中的所有调度器重新发送投票信息。在本发明实施例中，可在云计算系统的当前负载状况与主调度器不匹配时，重新选择主调度器，从而使得所选择的主调度器总是与云计算系统的负载状况相匹配。
第五方面，提供一种云计算系统，云计算系统包括仲裁器以及调度器；其中，调度器，用于确定云计算系统的当前负载状况，以及根据云计算系统的当前负载状况、预设的负载状况与投票信息的对应关系，确定投票信息以及发送投票信息至仲裁器；仲裁器，用于接收至少一个调度器发送的投票信息，以及根据至少一个调度器发送的投票信息，从多个调度器中选择一调度器作为主调度器，以使得主调度器按照所对应的调度策略对云计算系统中的计算资源进行调度，其中，所述主调度器所对应的所述调度策略与所述云计算系统的当前负载状况相匹配。第三方面、第四方面以及第五方面及各自的实现方式中更为具体的实现方法可与前述方面或实现方式相互参考。
由上可见，在本发明实施例中，首先调度器确定整个云计算系统的当前负载状况，然后，根据当前负载状况确定投票信息。仲裁器根据多个调度器各自的投票信息，从云计算系统的多个调度器中，选择与当前负载状况相匹配的调度器，作为主调度器，而主调度器将按照其所对应的调度策略对云计算系统中的计算资源进行调度。由此，通过仲裁器决策出主调度器，而整个云计算系统中除主调度外的辅调度器将不再对计算资源进行调度，从而可保证在需要调度器调度的时候整个云计算系统只有一个调度器在进行调度工作，进而可避免不同调度器间对计算资源的调度冲突问题。
由于根据各个调度器的投票信息是根据当前系统的负载状况而产生的，所选出的调度器能够适应当前系统的负载状况，从而提高了主调度器的针对性，增强了云计算系统 的调度性能。
此外，由于主调度器是根据各个调度器投票产生的，仲裁器仅需根据投票信息即可确定主调度器，而无需根据各个调度器的特性来进行选择，从而实现了仲裁器与各个调度器之间的解耦和，仲裁器对不同的多种调度器进行统一调度，且调度器可以灵活的进行变更和增减，提高了系统的灵活性。
附图说明
图1为本发明实施例所提供的云计算系统的一示意图；
图2为本发明实施例所提供的云计算系统的另一示意图；
图3为本发明实施例所提供的云计算系统的一示意图；
图4为本发明实施例所提供的确定云计算系统中主调度器的方法一示意图；
图5为本发明实施例所提供的云计算系统的一示意图；
图6为本发明实施例所提供的云计算系统的一示意图；
图7为本发明实施例所提供的云计算系统的一示意图；
图8为本发明实施例所提供的确定云计算系统中主调度器的装置的一示意图；
图9为本发明实施例所提供的确定云计算系统中主调度器的装置的一示意图；
图10为本发明实施例所提供的云计算系统的一示意图；
图11为本发明实施例所提供的仲裁器的一示意图；
图12为本发明实施例所提供的调度器的一示意图。
具体实施方式
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，并不是全部的实施例。
本发明提供一种从云计算系统中确定主调度器的方法及装置，如图1所示，该方法及装置的应用场景如下：
随着互联网的飞速发展，云计算应运而生；所谓云计算是一种计算模式，如图1所示，整个云计算系统主要由客户端、调度器和资源池组成；其中，调度器可接收客户端的请求，且根据客户端的请求，调度资源池中的计算资源给客户端使用。
目前，在图1所示的云计算系统中通常为单调度器，也就是一个云计算系统中只有一个调度器；但是随着云计算系统所承载业务的类型越来越多，云计算系统中的调度器也逐渐由单调度器演变为多调度器；其中，在多调度器的云计算系统中，每个调度器对应一种调度策略，用于对云计算系统中的计算资源进行调度。
由于在实际应用中，在多调度器的云计算系统中，不同调度器间是独立进行工作的，即不同调度器是独自对云计算系统中的计算资源进行调度的，因此，采用上述方法，可能会造成不同调度器所调度计算资源的冲突。
实施例一
基于上述应用场景，本发明提供一种从云计算系统中确定主调度器的方法，其中，该方法的核心思想为：首先确定整个云计算系统的当前负载状况，然后，从云计算系统 的多个调度器中，选择与当前负载状况相匹配的调度器，作为主调度器，而主调度器将按照其所对应的调度策略对云计算系统中的计算资源进行调度，而整个云计算系统中除主调度外的辅调度器将不再对计算资源进行调度，从而可保证在每个时刻，整个云计算系统只有一个调度器在进行调度工作，进而可避免不同调度器间对调度资源的冲突问题。
本发明首先提供一种云计算系统，如图3所示，该系统至少包括仲裁器、多个调度器以及资源池：
其中，资源池用于提供计算资源；每个调度器对应一种调度策略，用于按照其对应的调度策略，对资源池中的计算资源进行调度；仲裁器用于从多个调度器中，选择一调度器，作为主调度器；其中，被选为主调度的调度器，将按照其对应的调度策略对资源池中的计算资源进行调度，而未被选择为主调度器的调度器称为辅调度器，将停止对资源池中的计算资源进行调度；比如，整个云计算系统包括三个调度器，分别为调度器A、调度器B以及调度器C，其中，调度器A对应调度策略A，调度器B对应调度策略B，调度器C对应调度策略C；而此时，如果仲裁器将调度器A选为主调度器，那么调度器B和调度器C将为辅调度器，而在整个云计算系统中，调度器A将按照其对应的调度策略A对资源池中的计算资源进行调度，而调度器B以及调度器C将作为辅调度器，停止对资源池中的计算资源进行调度。
利用图3所示的云计算系统，如图4所示，本发明所提供的调度方法，具体如下：
步骤S41：调度器获取整个云计算系统中资源池的当前负载状况；
资源池的当前负载状况，是指当前资源池中各个服务器的各类资源的负载情况，当前负载状况可以体现为一些具体的参数，当参数满足特定条件(例如满足特定的阈值)时则认为资源池的负载状况处于某一特定的负载状况中；也可以通过监测资源池中是否出现了特定的触发事件，从而使得资源池满足某一负载状况的条件。在具体的实施例中，负载状况可以通过指资源池中所有服务器的资源总体负载状况，例如某一种或者多种资源在资源池中所有服务器的负载量的总量、或者占总资源量的比值的平均值或者加权平均值(可用以衡量资源的负载情况或者衡量某类资源是否为主要消耗的资源)；或者当前资源池中某一种或多种资源的负载量与其他资源负载量的差值(常用来衡量资源负载的均衡性)、等来衡量。在另一些具体的实施例中，负载状况也可以通过服务器中部分服务器是否触发特定的条件来进行判定是否为预设的某种负载状况，例如某些服务器的资源负载大于特定的阈值(可用以判断资源池中服务器的均衡负载情况)，或者某些服务器的某类资源占用超过了特定的阈值(可用于判断资源池中服务器的资源种类分配的均衡性)等。
资源池负载状况的获取，可以通过调度器主动进行采集，也可以通过其他云计算系统中的功能性节点采集后发送给调度器。资源负载状况的采集或者检测可以通过现有技术实践，例如根据各个服务器主动上报的资源负载信息进行确定，或者通过管理节点在任务分配时进行计算确定，或者通过监控服务器之间的数据流量进行估计等。
步骤S42：调度器根据云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系，确定投票信息；
在本发明实施例中，每个调度器内所预设的负载状况与投票信息的对应关系，可相同也可不同，相关技术人员，可根据需求，自行设置负载状况与投票信息的对应关系。
在本发明实施例中，所述投票信息为当前调度器为所述多个调度器中至少一个调度器的投票信息，所述投票信息可具体为被投票调度器的标识，或，被投票调度器的评分；比如，在本发明实施例中，所预设的负载状况与投票信息的对应关系为“负载1—调度器1，负载2—调度器2，以及负载3—调度器3”，那么，在云计算系统的当前负载为1时，当前调度器所匹配出的调度器为为“调度器1”，而“调度器1”具体为被投票调度器的标识，再如，所预设的负载状况与投票信息的对应关系为“负载1—调度器1评5分，负载2—调度器2评6分，负载3—调度器3评10分”，那么在整个云计算系统的当前负载为负载1时，调度器所匹配出的投票信息为“调度器1评5分”“而调度器1评5分”具体为被投票调度器的评分。
步骤S43：调度器上报投票信息至仲裁器；
步骤S44：仲裁器根据每个调度器的投票信息，从多个调度器中确定主调度器，其余调度器作为辅调度器，其中，主调度器将按照其对应的调度策略对资源池中的计算资源进行调度，而辅调度器将停止对资源池中的计算资源进行调度。
在本发明实施例中，可具体采用下述方法，从多个调度器中选择主调度器：
首先所述仲裁器根据每个调度器的投票信息，确定每个调度器的投票得分，然后，从多个调度器中，确定投票得分满足预设条件的调度器作为主调度器，而关于主调度器得分所满足的预设条件本领域技术人员可自行设置，比如可将预设条件设置为将得分最高的调度器作为主调度器，也可将预设条件设置为将得分最低的调度器作为主调度器，亦可将投票条件设置为将投票得分在一定区间的调度器作为主调度器，在此不再赘述。
在本发明实施例中，将以“预设条件为得分最高的调度器为主调度器”为例，详细介绍“选择主调度器”的过程：
第一种方式：这种方式可具体对应于投票信息为被投票调度器的标识，具体为：人工预先为每个调度器设置不同的优先级，然后，仲裁器在接收到每个调度器的投票信息后，根据发送投票信息的调度器的优先级为每个投票信息设置不同的权重；然后，根据每个调度器所获得的投票信息以及每个投票信息的权重，确定每个调度器的投票得分；最后，从多个调度器中，选择投票得分最高的调度器为主调度器；
比如，如图5所示，整个云计算系统，包括调度器A、B、C三个调度器，而预先将调度器A的优先级设置为高，调度器B的优先级设置为中，调度器C的优先级为低，且设置调度器A的投票权重为3，调度器B的投票权重为2，调度器C的投票权重为1；且调度器A投票给调度器B，调度器B投票给调度器C，而调度器C投票给调度器B，那么仲裁器在接收到上述每个调度器的投票信息后，将计算每个调度器的投票得分；其中，经计算可得出：调度器A的投票得分为0；调度器B的投票得分为：3*1+1*1＝4；调度器C的投票得分为：2*1＝2；可见，上述三个调度器中，调度器B的得分最高，因此仲裁器选择调度器B为主调度器。
第二种方式，这种方式可对应于投票信息为被投票调度器的评分，具体为：仲裁器直接统计每个调度器的评分，比如，仍沿用上述举例，整个云计算系统，包括调度器A、B、C三个调度器，那么，调度器A的投票信息为“调度器B评5分”，调度器B的投票信息为“调度器C评3分”，“调度器C的投票信息为”调度器C评6分”，那么，可统计调度器B的投票得分为5分，调度器C的投票得分为8分＝5分+3分，而调度器A的投票得分为0分。
第二种方式：这种方式可对应投票信息为被投票调度器的标识和被投票调度器的评分两种情况，具体为：每个调度器向仲裁器至少发送两个投票信息，其中一个投票信息为自己，表明当前调度器要参与选举，另一个投票为其它调度器，所述其它调度器为所述云计算系统中，除当前调度器外的剩余调度器；
而仲裁器在接收到调度器的投票信息后，首先判断当前调度器是否为其它调度器投票，如果为其它调度器投票，统计当前调度器的投票得分，否则，确定当前调度器的得分为零；
关于如何统计调度器的投票得分，可具体采用上述第一种方式所公开的方式，也可采用第二种普通的相加方式；
采用上述方式，在某一调度器只为自身投票时，仲裁器将该调度器的投票得分判为零，从而可避免某调度器只为自身投票的情况。
由上可见，在本发明实施例中，首先确定整个云计算系统的当前负载状况，然后，从云计算系统的多个调度器中，选择与当前负载状况相匹配的调度器，作为主调度器，而主调度器将按照其所对应的调度策略对云计算系统中的计算资源进行调度，而整个云计算系统中除主调度外的辅调度器将不再对计算资源进行调度，从而可保证在每个时刻，整个云计算系统只有一个调度器在进行调度工作，进而可避免不同调度器间对计算资源的调度冲突问题。
实施例二
在本发明实施例中，当采用实施例一所公开的方法，在整个云计算系统中选择出主调度器后，主调度器将按照其对应的调度策略对资源池中的计算资源进行调度；而未被选择为主调度器的辅调度器将实时监控整个云计算系统的负载状况，当负载状况不满足自身预设的条件时，向仲裁器上报重新投票信息；而仲裁器在发送重新投票信息的调度器的比例满足预设条件时，向所有调度器发送重新选举指令，而调度器将采用实施例一所公开的方式，重新发送投票信息，而仲裁器将重新选举出主调度器，从而使得所选举出的主调度器与云计算系统的当前负载状况相匹配。
实施例三
下面将以整个云计算系统，包括三个调度器、一个仲裁器和一个资源池为例，详细介绍，本发明的过程：
首先，如图6所示，在实际应用中，资源池可具体由多个服务器组成，每个服务器提供一定的计算资源，而上述三个调度器可具体为LB(Load Balanc，负载均衡)调度器、LC(Load Consolidation，负载整合)调度器以及HE(Hotspot Elimination，热点消除)调度器；其中，LB调度器所对应的调度策略为LB调度策略，LC调度器所对应的调度策略为LC调度策略，HE调度器所对应的调度策略为HE调度策略；
而上述LB调度策略，可具体为在对服务器进行业务分配时，尽量保证资源池中每个服务器的负载均衡；
LC调度策略，可具体为当资源池中的一服务器的负载低于预设最低值时，将该服务器的业务迁移到资源池中的其它服务器上，同时将该服务器下电；
HE调度策略，可具体为当资源池中某一服务器的负载超过预设最高值时，将该服务 器的部分业务迁移到资源池中的其它服务器上，从而降低当前服务器的负载。
在本发明实施例中，假设预先将HE调度器的优先级设置为高，将LB调度器的优先级设置为中，将LC调度器的优先级设置为低，而将高优先级HE调度器的投票信息的权重设置为3，将中优先级LB调度器的投票信息的权重设置为2，将低先级LC调度器的投票信息的权重设置为1；
在本发明实施例中，本领域技术人员，可自行设置每个调度器内部的投票策略；假设，在本发明实施例中，HE、LB和LC调度器的投票策略，各自如下：
HE调度器的投票策略：云计算系统的当前负载大于Th时，为HE投票；
                    云计算系统的当前负载小于T1时，为LC投票；
                    云计算系统的当前负载大于等于T1，小于等于Th时，为LB投票；
LB调度器的投票策略：云计算系统的当前负载小于T1时，为LC投票；
                    云计算系统的当前负载大于等于T1时，为LB投票；
LC调度器的投票策略：当云计算系统的当前负载小于T1时，并持续m分钟时，为LC投票；
                    而当云计算系统的当前负载大于等于T1时，不进行任何投票；
在本发明实施例中，假设整个云计算系统的当前负载状况小于T1，且持续了m分钟，那么通过上述投票策略，可确定HE调度器的投票信息为“为LC投票”，LB调度器的投票信息为“为LC投票”，LC调度器的投票信息为“为LC投票”；
那么仲裁器在接收到上述上述各个调度器的投票信息时，可确定LC调度器的投票得分为6＝3*1+2*1+1*1，而HE和LB调度器的投票得分均为0；此时仲裁器将LC调度器选为主调度器，HE和LB调度器为辅调度器，此时，LC调度器将按照上述LC调度策略对资源池中的计算资源进行调度，而HE和LB调度器将停止调度工作。
而在将LC调度器选为主调度器后，LC和LB调度器将实时监测云计算系统中当前负载的状况，而当云计算系统的当前负载不满足条件时，比如(当前负载大于Th时)，LB和LC调度器可向仲裁器发送重新选举的请求，而仲裁器在发送重新选举调度器的比例大于一定值时，触发所有调度器重新进行选举；比如，当发送重新选举调度器的比例占所述云系统中所有调度器的60％时，即重新进行选举。
由上可见，在本发明实施例中，采用上述方法，可保证在每个时刻，整个云计算系统中仅有一个调度器进行工作，从而可避免不同调度器间的调度冲突问题。
实施例四
本发明实施例，如图7所示，将以整个云计算系统包括三个调度器、一个仲裁器和一个资源池，且三个调度器分别为科学计算调度器、存储云调度器和Web服务调度器为例，详细介绍本发明的过程：
其中，科学计算调度器用于为科学计算业务，调度计算资源，其对应的调度策略为为科学计算业务提供充足的CPU资源，其对应的优先级为5；存储云调度器用于为存储云业务，调度计算资源，其对应的调度策略为尽量均衡磁盘输入输出和磁盘容量，减少碎片，其对应的优先级为3；Web服务调度器用于对Web业务，调度计算资源，其对应的调 度策略为为当前业务提供充足的CPU和网络资源，其对应的优先级为7；
在本发明实施例中，上述三个调度器可分别采集云计算系统的当前负载状况，所述云计算系统的当前负载可具体包括云计算系统的磁盘负载以及网络负载；
在本发明实施例中，上述三个调度器的调度策略可分别如下：
科学计算调度器：第一，为自己投票；第二，当资源池中的剩余磁盘资源高于Rs时，为存储云调度器投票，否则，为Web服务调度器投票；
存储云调度器：第一，为自己投票；第二，当资源池中的剩余剩网络资源低于Rw时，为Web服务调度器投票，否则，为科学计算调度器投票。
Web服务调度器：第一，为自已投票；第二，当资源池中的剩余网络资源高于Rs时，为存储云调度器投票，否则，为科学计算调度器投票。
首先，需要说明的是，在本发明实施例中，每个调度器可投两票，一票投自己，表明当前调度器参与选举，另一票，投给与当前负载状况相匹配的其它调度器。
在本发明实施例中，以云计算系统的当前负载状况下，云计算系统中的剩磁盘资源高于Rs，剩余网络资源低于Rw为例，详细介绍本发明的过程：
具体的，对应于上述每个调度器的投票策略，对于科学计算调度器的投票信息可具体为：为自己，即为科学计算调度器投票，为存储云调度器投票；而对于存储云调度器的投票信息可具体为：为自己投票，即为存储云调度器投票，为Web服务调度器投票；对于Web服务调度器的投票信息可具体为：为自己投票，即为Web服务调度器投票，为存储云调度器投票。
而所述仲裁器接收到每个调度器的投票信息时，仲裁器可具体首先判断当前调度器是否为自身投票，如果为自身投票，确定当前调度器有参选资格，否则，确定当前调度器无参选资格，不再统计当前调度器的得分；然后，再判断每个调度器是否为其它调度器投票，如果当前调度器为其它调度器投票，统计当前调度器的得分，否则，确定当前调度器的得分为零；采用上述方式，可避免优先级高的调度器仅为自身投票，而不为其它调度器投票，进而避免优先级高的调度器总被选举为主调度器。
在本发明实施例中，仍沿用上述举例，仲裁器可看到上述三个调度器均为自身投票，均有参选资源，且均为其它调度器进行投票，可正常统计每个调度器的得分。具体通过沿用实施例一中的计算方式，可得到科学计算调度器的投票得分为：5*1＝5分；存储云调度器的投票得分为：5*1+3*1+7*1＝15分；Web服务调度器的投票得分为：3*1+7*1＝10分；
在本发明实施例中，如果选择得分最高的调度器为主调度器，可看出，在本发明实施例中，将选举存储云调度器为主调度器，而科学计算调度器和Web服务调度器为辅调度器；而存储云调度器将按照其对应的调度策略，对资源池中的计算资源进行调度，而作为辅调度器的科学计算调度器和Web服务调度器可实时监控当前资源池的负载状况，所述负载状况包括磁盘资源状况和网络资源状况；而每个调度器一旦发现资源池的当前负载状况不满足自身的条件时，将向仲裁器发送重新选举的请求，而仲裁器在发送重新选举请求的调度器达到一定比例时，触发所有调度器重新进行选举。
由上可见，在本发明实施例中，采用上述方法，可保证在每个时刻，整个云计算系统中仅有一个调度器进行工作，从而可避免不同调度器间的调度冲突问题。
实施例五
与上述构思相同，本发明还提供一种从云计算系统中确定主调度器的装置，如图8所示，所述装置包括：
接收单元81，用于接收至少一个调度器发送的投票信息，所述投票信息为所述调度器根据所述云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系所生成的；
选择单元82，用于根据所述至少一个调度器发送的投票信息，从所述多个调度器中确定主调度器，以使得所述主调度器按照所对应的调度策略对所述云计算系统中的计算资源进行调度，其中，所述主调度器所对应的所述调度策略与所述云计算系统的当前负载状况相匹配。
可选的，选择单元，具体用于根据所述至少一个调度器发送的投票信息，确定每个调度器的投票得分；从所述多个调度器中，确定投票得分满足预设条件的调度器为所述主调度器。
可选的，预先为不同调度器设置不同的优先级，所述选择单元，在根据所述至少一个调度器发送的投票信息，确定每个调度器的投票得分时，具体用于：
根据每个调度器的优先级，为每个调度器的投票信息设置不同的权重；根据每个调度器所获得的投票信息以及每个投票信息的权重，确定每个调度器的投票得分。
可选的，所述选择单元，在根据所述至少一个调度器发送的投票信息，确定每个调度器的投票得分时，具体用于：针对一调度器，确定当前调度器是否为其它调度器投票，所述其它调度器为所述云计算系统中，除当前调度器外的剩余调度器；在确定当前调度器为其它调度器投票时，统计当前调度器的投票得分，否则，确定当前调度器的投票得分为零。
可选的，所述装置还包括，重新选择单元，具体用于：接收辅调度器发送的重新选举请求，所述辅调度器为所述云计算系统中，除所述主调度器外的其它调度器，所述重新选举请求为所述辅调度器在当前主调度器对云计算系统中计算资源的调度不满足预设条件时所发送的；在所述发送重新选举请求的辅调度器的比例满足预设条件时，向所有调度器发送重新选举指令，以使得所有调度器根据当前云计算系统的负载状况，重新进行投票。
由上可见，在本发明实施例中，首先确定整个云计算系统的当前负载状况，然后，从云计算系统的多个调度器中，选择与当前负载状况相匹配的调度器，作为主调度器，而主调度器将按照其所对应的调度策略对云计算系统中的计算资源进行调度，而整个云计算系统中除主调度外的辅调度器将不再对计算资源进行调度，从而可保证在每个时刻，整个云计算系统只有一个调度器在进行调度工作，进而可避免不同调度器间对计算资源的调度冲突问题。
实施例六
与上述构思相同，本发明还提供一种从云计算系统中确定主调度器的装置，如图9所示，所述装置包括：
负载确定单元91，用于确定所述云计算系统的当前负载状况；
投票信息确定单元92，用于根据所述云计算系统的当前负载状况以及预设的负载状 况与投票信息的对应关系，确定投票信息，其中，在所述云计算系统的当前负载状况下，所述投票信息所确定的投票结果对应的调度器为与当前负载状况所匹配的调度器；
发送单元93，用于发送所述投票信息至仲裁器，以使得所述仲裁器根据所述投票信息，确定主调度器，所述主调度器用于按照对应的调度策略对所述云计算系统中的计算资源进行调度。
可选的，所述装置还包括：采集单元，用于周期性采集云计算系统的负载状况；上报单元，用于在当前云计算系统的负载状况不满足预设条件时，上报
重新投票信息至所述仲裁器，以使得所述仲裁器根据所述重新投票信息，触发所述云计算系统中的所有调度器重新发送投票信息。
由上可见，在本发明实施例中，首先确定整个云计算系统的当前负载状况，然后，从云计算系统的多个调度器中，选择与当前负载状况相匹配的调度器，作为主调度器，而主调度器将按照其所对应的调度策略对云计算系统中的计算资源进行调度，而整个云计算系统中除主调度外的辅调度器将不再对计算资源进行调度，从而可保证在每个时刻，整个云计算系统只有一个调度器在进行调度工作，进而可避免不同调度器间对计算资源的调度冲突问题。
实施例七
与上述构思相同，本发明还提供一种云计算系统，如图10所示，所述云计算系统包括一仲裁器101以及多个调度器102；
其中，调度器102，用于确定所述云计算系统的当前负载状况，以及根据所述云计算系统的当前负载状况、预设的负载状况与投票信息的对应关系，确定投票信息以及发送所述投票信息至所述仲裁器；
仲裁器101，用于接收至少一个调度器发送的投票信息，以及根据所述至少一个调度器发送的投票信息，从多个调度器中选择一调度器作为主调度器，以使得所述主调度器按照所对应的调度策略对所述云计算系统中的计算资源进行调度，其中，所述主调度器所对应的所述调度策略与所述云计算系统的当前负载状况相匹配。
由上可见，在本发明实施例中，首先确定整个云计算系统的当前负载状况，然后，从云计算系统的多个调度器中，选择与当前负载状况相匹配的调度器，作为主调度器，而主调度器将按照其所对应的调度策略对云计算系统中的计算资源进行调度，而整个云计算系统中除主调度外的辅调度器将不再对计算资源进行调度，从而可保证在每个时刻，整个云计算系统只有一个调度器在进行调度工作，进而可避免不同调度器间对计算资源的调度冲突问题。
实施例八
与上述构思相同，本发明还提供一种仲裁器，如图11所示，所述仲裁器至少包括存储器111和处理器112；
存储器111，用于存储程序和指令；
处理器112，用于通过调用存储器中存储的程序和指令，执行：
接收至少一个调度器发送的投票信息，所述投票信息为所述调度器根据所述云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系所生成的；
根据所述至少一个调度器发送的投票信息，从多个调度器中确定主调度器，以使得所述主调度器按照所对应的调度策略对所述云计算系统中的计算资源进行调度，其中，所述主调度器所对应的所述调度策略与所述云计算系统的当前负载状况相匹配。
与上述构思相同，如图12所示，本发明还提供一种调度器，所述调度器至少包括存储器121和处理器122；
存储器121，用于存储程序和指令；
处理器122，用于通过调用存储器中存储的程序和指令，执行：
确定所述云计算系统的当前负载状况；
根据所述云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系，确定投票信息，其中，在所述云计算系统的当前负载状况下，所述多个调度器的投票信息所确定的调度器为与当前负载状况所匹配的调度器；
发送所述投票信息至仲裁器，以使得所述仲裁器根据所述投票信息，确定主调度器，所述主调度器用于按照对应的调度策略对所述云计算系统中的计算资源进行调度。
其中，总线架构可以包括任意数量的互联的总线和桥，具体由处理器代表的一个或多个处理器和存储器代表的存储器的各种电路链接在一起。总线架构还可以将诸如外围设备、稳压器和功率管理电路等之类的各种其他电路链接在一起，这些都是本领域所公知的，因此，本文不再对其进行进一步描述。总线接口提供接口。处理器负责管理总线架构和通常的处理，存储器可以存储处理器在执行操作时所使用的数据。
由上可见，在本发明实施例中，首先确定整个云计算系统的当前负载状况，然后，从云计算系统的多个调度器中，选择与当前负载状况相匹配的调度器，作为主调度器，而主调度器将按照其所对应的调度策略对云计算系统中的计算资源进行调度，而整个云计算系统中除主调度外的辅调度器将不再对计算资源进行调度，从而可保证在每个时刻，整个云计算系统只有一个调度器在进行调度工作，进而可避免不同调度器间对计算资源的调度冲突问题。
本领域内的技术人员应明白，本发明的实施例可提供为方法、系统、或计算机程序产品。因此，本发明可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本发明可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本发明是参照根据本发明实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
尽管已描述了本发明的优选实施例，但本领域内的技术人员一旦得知了基本创造性概念，则可对这些实施例作出另外的变更和修改。所以，所附权利要求意欲解释为包括优选实施例以及落入本发明范围的所有变更和修改。
显然，本领域的技术人员可以对本发明实施例进行各种改动和变型而不脱离本发明实施例的精神和范围。这样，倘若本发明实施例的这些修改和变型属于本发明权利要求及其等同技术的范围之内，则本发明也意图包含这些改动和变型在内。

Claims (15)

  1. 一种从云计算系统中确定主调度器的方法，其特征在于，所述云计算系统至少包括仲裁器和多个调度器，每个调度器对应一种调度策略，所述方法包括：
    所述仲裁器接收至少一个调度器发送的投票信息，所述投票信息为所述调度器根据所述云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系所生成的；
    所述仲裁器根据所述至少一个调度器发送的投票信息，从所述多个调度器中确定主调度器，以使得所述主调度器按照所对应的调度策略对所述云计算系统中的计算资源进行调度，其中，所述主调度器所对应的所述调度策略与所述云计算系统的当前负载状况相匹配。
  2. 根据权利要求1所述的方法，其特征在于，所述仲裁器根据所述至少一个调度器发送的投票信息，从所述多个调度器中确定主调度器，包括：
    所述仲裁器根据所述至少一个调度器发送的投票信息，确定每个调度器的投票得分；
    所述仲裁器从所述多个调度器中，确定投票得分满足预设条件的调度器为所述主调度器。
  3. 根据权利要求2所述方法，其特征在于，预先为不同调度器设置不同的优先级，所述仲裁器根据所述至少一个调度器发送的投票信息，确定每个调度器的投票得分，包括：
    所述仲裁器根据每个调度器的优先级，为每个调度器的投票信息设置不同的权重；
    所述仲裁器根据每个调度器所获得的投票信息以及每个投票信息的权重，确定每个调度器的投票得分。
  4. 根据权利要求2所述的方法，其特征在于，所述仲裁器根据所述至少一个调度器发送的投票信息，确定每个调度器的投票得分，包括：
    针对一调度器，所述仲裁器确定当前调度器是否为其它调度器投票，所述其它调度器为所述云计算系统中，除当前调度器外的剩余调度器；
    所述仲裁器在确定当前调度器为其它调度器投票时，统计当前调度器的投票得分，否则，确定当前调度器的投票得分为零。
  5. 根据权利要求1至4任一项所述的方法，其特征在于，所述方法还包括：
    所述仲裁器接收辅调度器发送的重新选举请求，所述辅调度器为所述云计算系统中，除所述主调度器外的其它调度器，所述重新选举请求为所述辅调度器在当前主调度器对云计算系统中计算资源的调度不满足预设条件时所发送的；
    所述仲裁器在所述发送重新选举请求的辅调度器的比例满足预设条件时，向所有调度器发送重新选举指令，以使得所有调度器根据当前云计算系统的负载状况，重新进行投票。
  6. 一种从云计算系统中确定主调度器的方法，其特征在于，所述云计算系统至少包括仲裁器和多个调度器，每个调度器对应一种调度策略，所述方法包括：
    所述调度器确定所述云计算系统的当前负载状况；
    所述调度器根据所述云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系，确定投票信息，其中，在所述云计算系统的当前负载状况下，所述多个调度 器的投票信息所确定的调度器为与当前负载状况所匹配的调度器；
    所述调度器发送所述投票信息至仲裁器，以使得所述仲裁器根据所述投票信息，确定主调度器，所述主调度器用于按照对应的调度策略对所述云计算系统中的计算资源进行调度。
  7. 根据权利要求6所述的方法，其特征在于，所述方法还包括：
    所述调度器周期性采集云计算系统的负载状况；
    所述调度器在当前云计算系统的负载状况不满足预设条件时，上报
    重新投票信息至所述仲裁器，以使得所述仲裁器根据所述重新投票信息，触发所述云计算系统中的所有调度器重新发送投票信息。
  8. 一种从云计算系统中确定主调度器的装置，其特征在于，所述云计算系统至少包括多个调度器，每个调度器对应一种调度策略，所述装置包括：
    接收单元，用于接收至少一个调度器发送的投票信息，所述投票信息为所述调度器根据所述云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系所生成的；
    选择单元，用于根据所述至少一个调度器发送的投票信息，从所述多个调度器中确定主调度器，以使得所述主调度器按照所对应的调度策略对所述云计算系统中的计算资源进行调度，其中，所述主调度器所对应的所述调度策略与所述云计算系统的当前负载状况相匹配。
  9. 根据权利要求8所述的装置，其特征在于，所述选择单元，具体用于
    根据所述至少一个调度器发送的投票信息，确定每个调度器的投票得分；
    从所述多个调度器中，确定投票得分满足预设条件的调度器为所述主调度器。
  10. 根据权利要求9所述的装置，其特征在于，预先为不同调度器设置不同的优先级，所述选择单元，在根据所述至少一个调度器发送的投票信息，确定每个调度器的投票得分时，具体用于：
    根据每个调度器的优先级，为每个调度器的投票信息设置不同的权重；
    根据每个调度器所获得的投票信息以及每个投票信息的权重，确定每个调度器的投票得分。
  11. 根据权利要求9所述的装置，其特征在于，所述选择单元，在根据所述至少一个调度器发送的投票信息，确定每个调度器的投票得分时，具体用于：
    针对一调度器，确定当前调度器是否为其它调度器投票，所述其它调度器为所述云计算系统中，除当前调度器外的剩余调度器；
    在确定当前调度器为其它调度器投票时，统计当前调度器的投票得分，否则，确定当前调度器的投票得分为零。
  12. 根据权利要求8至11任一项所述的装置，其特征在于，所述装置还包括，重新选择单元，具体用于：
    接收辅调度器发送的重新选举请求，所述辅调度器为所述云计算系统中，除所述主调度器外的其它调度器，所述重新选举请求为所述辅调度器在当前主调度器对云计算系统中计算资源的调度不满足预设条件时所发送的；
    在所述发送重新选举请求的辅调度器的比例满足预设条件时，向所有调度器发送重新选举指令，以使得所有调度器根据当前云计算系统的负载状况，重新进行投票。
  13. 一种从云计算系统中确定主调度器的装置，其特征在于，所述云计算系统至少包括仲裁器，所述装置包括：
    负载确定单元，用于确定所述云计算系统的当前负载状况；
    投票信息确定单元，用于根据所述云计算系统的当前负载状况以及预设的负载状况与投票信息的对应关系，确定投票信息，其中，在所述云计算系统的当前负载状况下，所述投票信息所确定的投票结果对应的调度器为与当前负载状况所匹配的调度器；
    发送单元，用于发送所述投票信息至仲裁器，以使得所述仲裁器根据所述投票信息，确定主调度器，所述主调度器用于按照对应的调度策略对所述云计算系统中的计算资源进行调度。
  14. 根据权利要求13所述的装置，其特征在于，所述装置还包括：
    采集单元，用于周期性采集云计算系统的负载状况；
    上报单元，用于在当前云计算系统的负载状况不满足预设条件时，上报
    重新投票信息至所述仲裁器，以使得所述仲裁器根据所述重新投票信息，触发所述云计算系统中的所有调度器重新发送投票信息。
  15. 一种云计算系统，其特征在于，所述云计算系统包括仲裁器以及调度器；
    其中，所述调度器，用于确定所述云计算系统的当前负载状况，以及根据所述云计算系统的当前负载状况、预设的负载状况与投票信息的对应关系，确定投票信息以及发送所述投票信息至所述仲裁器；
    所述仲裁器，用于接收至少一个调度器发送的投票信息，以及根据所述至少一个调度器发送的投票信息，从多个调度器中选择一调度器作为主调度器，以使得所述主调度器按照所对应的调度策略对所述云计算系统中的计算资源进行调度，其中，所述主调度器所对应的所述调度策略与所述云计算系统的当前负载状况相匹配。
PCT/CN2017/092854 2016-11-30 2017-07-13 从云计算系统中确定主调度器的方法及装置 WO2018099090A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP17877361.0A EP3541048A4 (en) 2016-11-30 2017-07-13 METHOD AND DEVICE FOR DETERMINING THE MAIN SCHEDULER OF A CLOUD COMPUTER SYSTEM
US16/425,680 US20190280945A1 (en) 2016-11-30 2019-05-29 Method and apparatus for determining primary scheduler from cloud computing system

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201611089191.9 2016-11-30
CN201611089191.9A CN108123987A (zh) 2016-11-30 2016-11-30 从云计算系统中确定主调度器的方法及装置

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/425,680 Continuation US20190280945A1 (en) 2016-11-30 2019-05-29 Method and apparatus for determining primary scheduler from cloud computing system

Publications (1)

Publication Number Publication Date
WO2018099090A1 true WO2018099090A1 (zh) 2018-06-07

Family

ID=62226476

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/092854 WO2018099090A1 (zh) 2016-11-30 2017-07-13 从云计算系统中确定主调度器的方法及装置

Country Status (4)

Country Link
US (1) US20190280945A1 (zh)
EP (1) EP3541048A4 (zh)
CN (1) CN108123987A (zh)
WO (1) WO2018099090A1 (zh)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108770014B (zh) * 2018-07-10 2022-03-25 广东工业大学 网络服务器的计算评估方法、系统、装置及可读存储介质
CN109917705B (zh) * 2019-02-25 2021-10-22 弗徕威智能机器人科技（上海）有限公司 一种多任务调度方法
CN110737521B (zh) * 2019-10-14 2021-03-05 中国人民解放军32039部队 基于任务调度中心的容灾方法和装置
US11609811B2 (en) * 2020-08-27 2023-03-21 Microsoft Technology Licensing, Llc Automatic root cause analysis and prediction for a large dynamic process execution system
CN112527512B (zh) * 2020-12-31 2023-11-21 河南信大网御科技有限公司 一种增强型拟态构造系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102387173A (zh) * 2010-09-01 2012-03-21 中国移动通信集团公司 一种MapReduce系统及其调度任务的方法和装置
CN103297499A (zh) * 2013-04-19 2013-09-11 无锡成电科大科技发展有限公司 一种基于云平台的调度方法及系统
US20150026306A1 (en) * 2013-07-16 2015-01-22 Electronics And Telecommunications Research Institute Method and apparatus for providing virtual desktop service
CN104657221A (zh) * 2015-03-12 2015-05-27 广东石油化工学院 一种云计算中基于任务分类的多队列错峰调度模型及方法
US20150324231A1 (en) * 2009-10-26 2015-11-12 Microsoft Technology Licensing, Llc Opportunistically scheduling and adjusting time slices
CN106020941A (zh) * 2015-03-24 2016-10-12 国际商业机器公司 选择资源分配策略和解决资源冲突

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2012034273A1 (en) * 2010-09-15 2012-03-22 Empire Technology Development Llc Task assignment in cloud computing environment
US9658874B2 (en) * 2014-06-18 2017-05-23 International Business Machines Corporation Selecting optimal hypervisor platforms that satisfy application workload requirements
CN105634784B (zh) * 2014-11-06 2019-02-05 阿里巴巴集团控股有限公司 控制数据分发方法、装置及系统
CN105791371B (zh) * 2014-12-22 2019-01-04 中国科学院上海高等研究院 一种云存储服务系统及方法
CN106155780B (zh) * 2015-04-02 2020-01-31 阿里巴巴集团控股有限公司 一种基于时间的节点选举方法及装置
US10333853B1 (en) * 2016-03-30 2019-06-25 Amazon Technologies, Inc. Unified quality of service (QoS) for label switching traffic
CN105743995B (zh) * 2016-04-05 2019-10-18 北京轻元科技有限公司 一种可移植高可用部署和管理容器集群的系统和方法
US10552228B2 (en) * 2016-06-20 2020-02-04 Vmware, Inc. Distributed hierarchical scheduler

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150324231A1 (en) * 2009-10-26 2015-11-12 Microsoft Technology Licensing, Llc Opportunistically scheduling and adjusting time slices
CN102387173A (zh) * 2010-09-01 2012-03-21 中国移动通信集团公司 一种MapReduce系统及其调度任务的方法和装置
CN103297499A (zh) * 2013-04-19 2013-09-11 无锡成电科大科技发展有限公司 一种基于云平台的调度方法及系统
US20150026306A1 (en) * 2013-07-16 2015-01-22 Electronics And Telecommunications Research Institute Method and apparatus for providing virtual desktop service
CN104657221A (zh) * 2015-03-12 2015-05-27 广东石油化工学院 一种云计算中基于任务分类的多队列错峰调度模型及方法
CN106020941A (zh) * 2015-03-24 2016-10-12 国际商业机器公司 选择资源分配策略和解决资源冲突

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3541048A4

Also Published As

Publication number Publication date
EP3541048A1 (en) 2019-09-18
CN108123987A (zh) 2018-06-05
EP3541048A4 (en) 2019-11-20
US20190280945A1 (en) 2019-09-12

Similar Documents

Publication Publication Date Title
WO2018099090A1 (zh) 从云计算系统中确定主调度器的方法及装置
CN106776005B (zh) 一种面向容器化应用的资源管理系统及方法
US8745216B2 (en) Systems and methods for monitoring and controlling a service level agreement
JP6731201B2 (ja) 時間ベースのノード選出方法及び装置
CN106375420B (zh) 一种基于负载均衡的服务器集群智能监控系统及方法
US9875135B2 (en) Utility-optimized scheduling of time-sensitive tasks in a resource-constrained environment
WO2015062492A1 (zh) 一种业务处理方法、系统及设备
US9652150B2 (en) Global memory sharing method and apparatus, and communications system
US8533731B2 (en) Apparatus and method for distrubuting complex events based on correlations therebetween
CN111950988B (zh) 分布式工作流调度方法、装置、存储介质及电子设备
CN102667724A (zh) 利用加速器的、目标指向的工作负载性能管理
US8375228B2 (en) Multiple-node system power utilization management
US20130067267A1 (en) Resource aware placement of applications in clusters
US7600229B1 (en) Methods and apparatus for load balancing processing of management information
US20180165112A1 (en) Managing test services in a distributed production service environment
US20220327002A1 (en) Allocating computing resources for deferrable virtual machines
CN112015549B (zh) 一种基于服务器集群的调度节点的选择抢占方法及系统
CN109412838A (zh) 基于散列计算以及性能评估的服务器集群主节点选择方法
US8510273B2 (en) System, method, and computer-readable medium to facilitate application of arrival rate qualifications to missed throughput server level goals
CN113391911A (zh) 一种大数据资源动态调度方法、装置和设备
CN111488373A (zh) 用于处理请求的方法和系统
KR101526344B1 (ko) 유사 이메일을 처리하기 위한 시스템 및 방법
CN115168042A (zh) 监控集群的管理方法及装置、计算机存储介质、电子设备
CN114864069A (zh) 用于调度任务的方法及装置
Khalifa et al. MobiCloud: A reliable collaborative mobilecloud management system

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17877361

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2017877361

Country of ref document: EP

Effective date: 20190614