CN106790636A - A kind of equally loaded system and method for cloud computing server cluster - Google Patents
A kind of equally loaded system and method for cloud computing server cluster Download PDFInfo
- Publication number
- CN106790636A CN106790636A CN201710013263.XA CN201710013263A CN106790636A CN 106790636 A CN106790636 A CN 106790636A CN 201710013263 A CN201710013263 A CN 201710013263A CN 106790636 A CN106790636 A CN 106790636A
- Authority
- CN
- China
- Prior art keywords
- user
- module
- gpu
- gpu calculation
- server
- 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
Classifications
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Computer And Data Communications (AREA)
Abstract
The invention discloses a kind of equally loaded system and method for cloud computing server cluster, using two layers of resource allocation policy, specifically:During User logs in, cooperated by the login module in management server, monitoring module, distribute module, assign them to a certain GPU calculation server, when user submits operation on GPU calculation servers, the resource of management server estimates the occupation condition that module will estimate operation, with reference to the GPU calculation server real time resources occupancy situations that monitoring module is monitored, such as find that allocated inadequate resource to complete the task, then carries out the second Layer assignment by distribute module.The present invention is based on event-driven mechanism, realize the load balancing of cluster, with automation, it is intelligent, perform online the characteristics of, with user experience it is main to consider when load balancing is carried out, so that load balancing is simpler, efficient, transparent, it is ensured that the high-performance of group system, high quality-of-service and resilient expansion.
Description
Technical field
The present invention relates to cluster server technical field, and in particular to a kind of equally loaded system of cloud computing server cluster
System and method.
Background technology
Along with the development of the demand of calculating, the development of cloud computing is more and more faster.Cloud computing cluster is generally using extensive meter
Calculate module to be calculated, in order to ensure the experience of security, stability and the user of cluster, it is necessary to using load balancing collection
Group.
The drawbacks of conventional cluster load balance system typically can only carry out primary distribution, this mechanism to resource at present exists
In after distribution when user really submits task to and starts computing, the server resource according to rule distribution differs and surely meet
The calculating demand of user, and because distribution is individual layer, unidirectional, static state, so easily cause server resource overload, user making
Occur with not good situation is experienced.
The problems such as in view of cost performance, Consumer's Experience, therefore, realize the automation of cluster load, intellectuality, online
Change is problem demanding prompt solution.
The content of the invention
It is an object of the invention to provide a kind of equally loaded system and method for cloud computing server cluster, it is used to solve
Existing cluster load balance system is not enough automated, intelligent, onlineization, cost performance and the not good problem of Consumer's Experience.
To achieve the above object, the present invention devises a kind of equally loaded system and method for cloud computing server cluster.
Specifically, a kind of equally loaded system of cloud computing server cluster, it is important to, the system includes:For depositing user data
With user calculate data storage server, one group be used for perform user's calculating task GPU calculation servers and for distributing,
Monitor the management server of GPU calculation server ruuning situations;
Wherein, described management server includes:
Login module:For obtaining and preserving user login information, and user login information is transmitted to monitoring module and is divided
With module;
Monitoring module:Number of users, GPU occupancy situations, EMS memory occupation feelings for monitoring each GPU calculation server
Condition;
Distribute module:For the user of login to be distributed into specific calculation server;
Resource estimates module:Resource for being taken after user submits operation to is estimated.
Further, the maximum of concurrent user's quantity that described GPU calculation servers can be accommodated is set by keeper.
Further, public directory and User Catalog are provided with described storage server, User Catalog includes that institute is useful
The respective catalogue in family, can call the information and data in own directory after User logs in.
A kind of equally loaded method of the cloud computing server cluster based on said system, methods described includes following step
Suddenly:
When a, User logs in, the log-on message of user is obtained by the login module in management server, then triggering is logged in
Event, after log-in events triggering, monitoring module is immediately by the number of users of all GPU calculation servers, GPU occupancy situations, interior
Deposit occupancy situation and be sent to distribute module, user is distributed to the minimum GPU calculation servers of active user's number by distribute module;
B, user call the data in storage server, in the GPU calculation servers operation that management server is its distribution
Task, when submitting task to, triggering resource estimates event, when estimating module according to the task of submission to by the resource in management server
Description and selection parameter are estimated to the resource that task takes, and described predictor method includes according to data type, size, appoints
Software used in business flow, finds estimated needs in the database that the experience accumulated according to conventional operation task is created
GPU occupancies, and internal memory, IO, the network capacity for needing, the real time resources transmitted by estimation results and monitoring module are taken
Situation, if finding, allocated inadequate resource, to complete task, sub-distribution again is carried out by distribute module;
After c, task run are finished, user logs off, and terminates.
Further, in described step a, user is distributed to the minimum GPU of active user's number and calculates clothes by distribute module
During business device, if the minimum GPU calculation servers quantity of active user's number is more than 1, user is sequentially allocated to GPU and calculates clothes
Business device.
Further, in step a, after User logs in, the number in public directory and User Catalog in storage server is called
According to.
Further, in described step a, user is distributed to the minimum GPU of active user's number and calculates clothes by distribute module
During business device, if all reaching maximum concurrent user number when monitoring module monitors all GPU calculation servers, sent out to login module
Deliver letters breath, login module issues the user with prompt message.
Further, in described step a, user is distributed to the minimum GPU of active user's number and calculates clothes by distribute module
During business device, the 80% of maximum concurrent user number is all reached when monitoring module monitors all GPU calculation servers, then to keeper
Send prompt message.
Further, keeper has number of users, GPU occupancy situations, the internal memory of the GPU calculation servers in checking monitoring module
The authority of the information of occupancy situation, keeper has stopping consumer process volume authority, and users personal data looks into storage server
See authority only individual subscriber.
Further, be stored in the log-on message of user in database by the login module in management server, step c,
User can select temporary transient offline or end process when logging off, if selection is temporarily offline, log-on message is not from data
Deleted in storehouse, system thinks the user or logging status, when user logs in from webpage or client again, will be automatically matched to
GPU calculation servers before, the process before continuation;If selection end process, log-on message is by from the data of login module
Deleted in storehouse, user logs in and will redistribute GPU calculation servers next time.
The invention has the advantages that:1st, cloud computing platform is often that, for certain specific area, this just determines use
Task be concentrated mainly in a limited scope, this carries out resource and estimates when making it possible that user submits task to.This
Scheme utilizes resource pre-estimating technology, carries out double-deck resource allocation, makes every effort to reach the dynamic, abundant, effective of the utilization of resources.2nd, this hair
It is bright that the load balancing of cluster is realized based on event driven double-deck distribution mechanism, with automation, intelligent, online perform
Feature, with user experience is main to consider when load balancing is carried out so that load balancing is simpler, efficient,
It is transparent, it is ensured that the high-performance of group system, high quality-of-service and resilient expansion.
Brief description of the drawings
Fig. 1 is system architecture diagram of the invention.
Fig. 2 is flow chart of the method for the present invention.
Specific embodiment
Following examples are used to illustrate the present invention, but are not limited to the scope of the present invention.
The technical scheme deployment including system physical structure first, for details, reference can be made to Fig. 1, the cloud computing service
The physical arrangement of the equally loaded system of device cluster includes:
Management server, for distributing user, monitoring calculation server ruuning situation,
GPU calculation servers, the calculating task for performing user,
Storage server, for depositing user data and calculating data.
Above-mentioned management server includes:
Login module, for obtaining and preserving user login information, and is transmitted to user login information monitoring module and divides
With module;
Monitoring module, number of users, GPU occupancy situations, EMS memory occupation situation for monitoring each calculation server;
Distribute module, for the user of login to be distributed into specific calculation server;
Resource estimates module, for after user submits operation to carrying out that estimating for resource may be taken.
Wherein, heretofore described GPU calculation servers can have multiple, and each GPU calculation servers most multipotency is accommodated
Concurrent user number defined by keeper.
Further, there are public directory and User Catalog under storage server of the present invention, User Catalog includes that institute is useful
The respective catalogue in family.
Referring to Fig. 2, user uses a kind of flow of the equally loaded system of cloud computing server cluster of the present invention
It is as follows:
A) user initiates to log in from webpage or client;
B) login module in management server obtains the log-on message of user, then triggers log-in events;
C) after log-in events triggering, monitoring module in management server by the number of users of all GPU calculation servers,
GPU occupancy situations, EMS memory occupation situation are sent to the distribute module in management server;
D) distribute module carries out the first Layer assignment, and user is distributed into the minimum GPU calculation servers of active user's number, if
The minimum GPU calculation servers quantity of active user's number is more than 1, then user is sequentially allocated into GPU calculation servers;
E) user submits to task, triggering resource to estimate event on the GPU calculation servers being assigned to;
The parameter of description and selection when f) resource estimates module according to user's submission task carries out resource allocation and estimates;Institute
The predictor method stated is included according to software used in data type, size, flow of task, what is accumulated according to conventional operation task
The GPU occupancies of estimated needs, and internal memory, IO, the network capacity for needing are found in the database that experience is created;
G) distribute module is carried out according to the occupation condition estimated and the occupancy situation of current each GPU calculation servers
Second Layer assignment, user is distributed to can to greatest extent meet the GPU calculation servers of its use demand;
H) user's operation task;
I) user can call the data in public directory and oneself catalogue in storage server, and can by data be stored in from
In oneself catalogue;
J) when user submits task to every time, all trigger resource and estimate event, carry out two Layer assignments;
K) user exits, and can select temporary transient offline or end process, if selection is temporarily offline, log-on message is not from number
Deleted according in storehouse, system thinks the user or logging status, when user logs in from webpage or client again, by Auto-matching
GPU calculation servers before, the process before continuation;If selection end process, log-on message is by from the number of login module
Deleted according in storehouse, user logs in and will redistribute GPU calculation servers next time.
Wherein, in the method flow, when log-in events are triggered, if monitoring module monitors all GPU calculation servers all
Maximum concurrent user number is reached, then sends information to login module, login module gives the user that " login user number has reached
Limit, keeper please be contact " prompting.
If monitoring module monitors all GPU calculation servers and all reaches the 80% of maximum concurrent user number, system to
Keeper provides the prompting of " login user is more, please increase GPU calculation servers ".
Keeper with checking monitoring module, can obtain the number of users of all GPU calculation servers, CPU occupancy situations, interior
The information of occupancy situation is deposited, but does not check the authority of users personal data in storage server, keeper can also stop
The process of certain user.
Embodiment 1
Monitoring module in step 1, management server by receive number of users that GPU calculation servers periodically send,
The information such as GPU occupancy situations, EMS memory occupation situation are monitored to GPU calculation servers, if the use of all GPU calculation servers
Amount reaches the 80% of maximum concurrent user number, then perform step 2;If the number of users of all GPU calculation servers reaches
Maximum concurrent user number, then perform step 3;Otherwise, then step 4 is performed;
Step 2, system provide the prompting of " login user is more, please increase GPU calculation servers ", keeper to keeper
Increase a GPU calculation server S n+1 online, skip to step 4;
Step 3, user initiate to log in from webpage or client, and login module gives the user that " login user number has reached
The upper limit, please contact keeper " prompting, skip to step 14;
Step 4, user initiate to log in from webpage or client, trigger log-in events, and login module believes the login of user
Breath is sent to monitoring module, and the log-on message of user is saved in database;
Step 5, monitoring module are by the number of users of all GPU calculation servers, GPU occupancy situations, EMS memory occupation situation
It is sent to the distribute module in management server;
If the minimum GPU calculation servers quantity of step 6, active user's number is equal to 1, step 7 is performed;If active user
The minimum GPU calculation servers quantity of number is more than 1, then perform step 8;
User is distributed to the minimum GPU calculation servers of the number of users by step 7, distribute module, if keeper performs
Increase the operation of GPU calculation server S n+1 in step 2, then in the method, user is first distributed into S n+1, until S n+
Untill 1 number of users is identical with other GPU calculation server numbers of users, step 9 is skipped to;
Step 8, by user according to S1, S2, S3 ... Sn's, Sn+1 is sequentially assigned to GPU calculation servers;
Step 9, user submit to task, triggering resource to estimate event, resource estimate module according to the description of the task of submission to and
Parameter carries out resource and estimates, if the GPU calculation servers of estimated current distribution disclosure satisfy that the calculating demand, performs step
11, otherwise perform step 10;
Step 10, distribute module estimate occupation condition that module estimates according to resource and that monitoring module is observed is current
The occupation condition of all GPU calculation servers carries out second layer allocation schedule;
Step 11, user carry out computing on the GPU calculation servers of distribution, can call public mesh in storage server
Data in record and oneself catalogue, and data can be stored in the catalogue of oneself;
Step 12, user exit, and can select temporary transient offline or end process, if selection is temporarily offline, perform step
13;If selection end process, performs step 14;
Step 13, log-on message are deleted not from database, and system thinks the user or logging status, user again from
When webpage or client are logged in, by the GPU calculation servers before being automatically matched to, the process before continuation;
Step 14, log-on message will be deleted from the database of login module, and user logs in and will be opened from step 1 again next time
Begin, distribute GPU calculation servers;
Step 15, logs off, and terminates.
Step 1 is always in commission in the present invention.The present invention is based on event driven double-deck distribution mechanism, realizes cluster
Load balancing, with automation, it is intelligent, perform online the characteristics of, when load balancing is carried out based on user experience
Consider so that load balancing is simpler, efficient, transparent, it is ensured that the high-performance of group system, high quality-of-service and
Resilient expansion.
Although the present invention is described in detail above to have used general explanation and specific embodiment, at this
On the basis of invention, it can be made some modifications or improvements, this will be apparent to those skilled in the art.Therefore,
These modifications or improvements, belong to the scope of protection of present invention without departing from theon the basis of the spirit of the present invention.
Claims (10)
1. a kind of equally loaded system of cloud computing server cluster, it is characterised in that the system includes:For depositing number of users
The storage server of data, one group of GPU calculation server for being used to perform user's calculating task are calculated according to user and for dividing
With, monitoring GPU calculation server ruuning situations management server;
Wherein, described management server includes:
Login module:For obtaining and preserving user login information, and user login information is transmitted to monitoring module and distribution mould
Block;
Monitoring module:For monitor the number of users of each GPU calculation server, GPU occupancy situations, EMS memory occupation situation and
IO, network capacity;
Distribute module:For the user of login to be distributed into specific calculation server;
Resource estimates module:Resource for being taken after user submits operation to is estimated.
2. the equally loaded system of a kind of cloud computing server cluster according to claim 1, it is characterised in that described
The maximum of concurrent user's quantity that GPU calculation servers can be accommodated is set by keeper.
3. the equally loaded system of a kind of cloud computing server cluster according to claim 1, it is characterised in that described
Public directory and User Catalog are provided with storage server, User Catalog includes the respective catalogue of all users, after User logs in
The information and data in own directory can be called.
4. a kind of equally loaded method of cloud computing server cluster of system according to claim 1, it is characterised in that institute
The method of stating is comprised the following steps:
When a, User logs in, the log-on message of user is obtained by the login module in management server, then triggers log-in events,
After log-in events triggering, monitoring module is immediately by the number of users of all GPU calculation servers, GPU occupancy situations, EMS memory occupation
Situation is sent to distribute module, and user is distributed to the minimum GPU calculation servers of active user's number by distribute module;
B, user call the data in storage server, in the GPU calculation server operation tasks that management server is its distribution,
During submission task, triggering resource estimates event, description when estimating module according to the task of submission to by the resource in management server
The resource that task takes is estimated with selection parameter, described predictor method is included according to data type, size, task flow
Software used in journey, the GPU that estimated needs are found in the database that the experience accumulated according to conventional operation task is created is accounted for
With rate, and internal memory, IO, the network capacity for needing, the real time resources occupancy situation transmitted by estimation results and monitoring module,
If it was found that allocated inadequate resource carries out sub-distribution again to complete task by distribute module;
After c, task run are finished, user logs off, and terminates.
5. the equally loaded method of a kind of cloud computing server cluster according to claim 4, it is characterised in that described
In step a, when user is distributed to the minimum GPU calculation servers of active user's number by distribute module, if active user's number is minimum
GPU calculation servers quantity be more than 1 when, user is sequentially allocated to GPU calculation servers.
6. a kind of equally loaded method of cloud computing server cluster according to claim 4, it is characterised in that step a
In, after User logs in, call the data in public directory and User Catalog in storage server.
7. the equally loaded method of a kind of cloud computing server cluster according to claim 4, it is characterised in that described
In step a, when user is distributed to the minimum GPU calculation servers of active user's number by distribute module, if when monitoring module monitoring
Maximum concurrent user number is all reached to all GPU calculation servers, then sends information to login module, login module is sent out to user
Go out prompt message.
8. the equally loaded method of a kind of cloud computing server cluster according to claim 4, it is characterised in that described
In step a, when user is distributed to the minimum GPU calculation servers of active user's number by distribute module, when monitoring module is monitored
All GPU calculation servers all reach the 80% of maximum concurrent user number, then send prompt message to keeper.
9. a kind of equally loaded method of cloud computing server cluster according to claim 8, it is characterised in that keeper
There are number of users, GPU occupancy situations, the power of the information of EMS memory occupation situation of the GPU calculation servers in checking monitoring module
Limit, keeper has stopping consumer process volume authority, and users personal data checks authority only individual subscriber in storage server.
10. a kind of equally loaded method of cloud computing server cluster according to claim 4, it is characterised in that management
Be stored in the log-on message of user in database by the login module in server, and step c, user can select when logging off
Temporary transient offline or end process is selected, if selection is temporarily offline, log-on message is deleted not from database, and system thinks the user
Or logging status, when user logs in from webpage or client again, by the GPU calculation servers before being automatically matched to, after
Process before continuous;If selection end process, log-on message will be deleted from the database of login module, and user logs in next time will
Redistribute GPU calculation servers.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710013263.XA CN106790636A (en) | 2017-01-09 | 2017-01-09 | A kind of equally loaded system and method for cloud computing server cluster |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710013263.XA CN106790636A (en) | 2017-01-09 | 2017-01-09 | A kind of equally loaded system and method for cloud computing server cluster |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106790636A true CN106790636A (en) | 2017-05-31 |
Family
ID=58950733
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710013263.XA Pending CN106790636A (en) | 2017-01-09 | 2017-01-09 | A kind of equally loaded system and method for cloud computing server cluster |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106790636A (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107391627A (en) * | 2017-06-30 | 2017-11-24 | 北京奇虎科技有限公司 | EMS memory occupation analysis method, device and the server of data |
CN107391257A (en) * | 2017-06-30 | 2017-11-24 | 北京奇虎科技有限公司 | Predictor method, device and the server of memory size needed for business |
CN107463642A (en) * | 2017-07-19 | 2017-12-12 | 北京京东尚科信息技术有限公司 | The method and apparatus for lifting Tool for Data Warehouse resource utilization |
CN108595367A (en) * | 2018-04-25 | 2018-09-28 | 银川华联达科技有限公司 | A kind of server system based on computer cluster in LAN |
CN109165093A (en) * | 2018-07-31 | 2019-01-08 | 宁波积幂信息科技有限公司 | A kind of calculate node cluster elasticity distribution system and method |
CN109271251A (en) * | 2018-08-09 | 2019-01-25 | 深圳市瑞云科技有限公司 | A method of by constraining come scheduling node machine |
CN109561054A (en) * | 2017-09-26 | 2019-04-02 | 华为技术有限公司 | A kind of data transmission method, controller and access device |
CN110019110A (en) * | 2017-07-28 | 2019-07-16 | 腾讯科技(深圳)有限公司 | A kind of capacity management methods of operation system, device, equipment and operation system |
CN111552550A (en) * | 2020-04-26 | 2020-08-18 | 星环信息科技(上海)有限公司 | Task scheduling method, device and medium based on GPU (graphics processing Unit) resources |
CN111679911A (en) * | 2020-06-04 | 2020-09-18 | 中国建设银行股份有限公司 | Management method, device, equipment and medium for GPU (graphics processing Unit) card in cloud environment |
CN112685170A (en) * | 2019-10-18 | 2021-04-20 | 伊姆西Ip控股有限责任公司 | Dynamic optimization of backup strategies |
CN113806011A (en) * | 2021-08-17 | 2021-12-17 | 曙光信息产业股份有限公司 | Cluster resource control method and device, cluster and computer readable storage medium |
CN116112497A (en) * | 2022-12-29 | 2023-05-12 | 天翼云科技有限公司 | Node scheduling method, device, equipment and medium of cloud host cluster |
Citations (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101753832A (en) * | 2008-12-04 | 2010-06-23 | 北京中星微电子有限公司 | Cloud mirror control method in video monitoring system, system and central platform server |
CN102185779A (en) * | 2011-05-11 | 2011-09-14 | 田文洪 | Method and device for realizing data center resource load balance in proportion to comprehensive allocation capability |
CN102223419A (en) * | 2011-07-05 | 2011-10-19 | 北京邮电大学 | Virtual resource dynamic feedback balanced allocation mechanism for network operation system |
CN102333106A (en) * | 2010-07-13 | 2012-01-25 | 中国移动通信集团公司 | Peer-to-peer (P2P) system resource scheduling method and device and system thereof |
CN102594869A (en) * | 2011-12-30 | 2012-07-18 | 深圳市同洲视讯传媒有限公司 | Method and device for dynamically distributing resources under cloud computing environment |
CN103257896A (en) * | 2013-01-31 | 2013-08-21 | 南京理工大学连云港研究院 | Max-D job scheduling method under cloud environment |
CN103729246A (en) * | 2013-12-31 | 2014-04-16 | 浪潮(北京)电子信息产业有限公司 | Method and device for dispatching tasks |
CN103763378A (en) * | 2014-01-24 | 2014-04-30 | 中国联合网络通信集团有限公司 | Task processing method and system and nodes based on distributive type calculation system |
CN103823718A (en) * | 2014-02-24 | 2014-05-28 | 南京邮电大学 | Resource allocation method oriented to green cloud computing |
CN104202388A (en) * | 2014-08-27 | 2014-12-10 | 福建富士通信息软件有限公司 | Automatic load balancing system based on cloud platform |
CN104794194A (en) * | 2015-04-17 | 2015-07-22 | 同济大学 | Distributed heterogeneous parallel computing system facing large-scale multimedia retrieval |
CN104951372A (en) * | 2015-06-16 | 2015-09-30 | 北京工业大学 | Method for dynamic allocation of Map/Reduce data processing platform memory resources based on prediction |
CN106027318A (en) * | 2016-07-24 | 2016-10-12 | 成都育芽科技有限公司 | Cloud computing-based two-level optimal scheduling management platform for virtual machine |
US20160323377A1 (en) * | 2015-05-01 | 2016-11-03 | Amazon Technologies, Inc. | Automatic scaling of resource instance groups within compute clusters |
-
2017
- 2017-01-09 CN CN201710013263.XA patent/CN106790636A/en active Pending
Patent Citations (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101753832A (en) * | 2008-12-04 | 2010-06-23 | 北京中星微电子有限公司 | Cloud mirror control method in video monitoring system, system and central platform server |
CN102333106A (en) * | 2010-07-13 | 2012-01-25 | 中国移动通信集团公司 | Peer-to-peer (P2P) system resource scheduling method and device and system thereof |
CN102185779A (en) * | 2011-05-11 | 2011-09-14 | 田文洪 | Method and device for realizing data center resource load balance in proportion to comprehensive allocation capability |
CN102223419A (en) * | 2011-07-05 | 2011-10-19 | 北京邮电大学 | Virtual resource dynamic feedback balanced allocation mechanism for network operation system |
CN102594869A (en) * | 2011-12-30 | 2012-07-18 | 深圳市同洲视讯传媒有限公司 | Method and device for dynamically distributing resources under cloud computing environment |
CN103257896A (en) * | 2013-01-31 | 2013-08-21 | 南京理工大学连云港研究院 | Max-D job scheduling method under cloud environment |
CN103729246A (en) * | 2013-12-31 | 2014-04-16 | 浪潮(北京)电子信息产业有限公司 | Method and device for dispatching tasks |
CN103763378A (en) * | 2014-01-24 | 2014-04-30 | 中国联合网络通信集团有限公司 | Task processing method and system and nodes based on distributive type calculation system |
CN103823718A (en) * | 2014-02-24 | 2014-05-28 | 南京邮电大学 | Resource allocation method oriented to green cloud computing |
CN104202388A (en) * | 2014-08-27 | 2014-12-10 | 福建富士通信息软件有限公司 | Automatic load balancing system based on cloud platform |
CN104794194A (en) * | 2015-04-17 | 2015-07-22 | 同济大学 | Distributed heterogeneous parallel computing system facing large-scale multimedia retrieval |
US20160323377A1 (en) * | 2015-05-01 | 2016-11-03 | Amazon Technologies, Inc. | Automatic scaling of resource instance groups within compute clusters |
CN104951372A (en) * | 2015-06-16 | 2015-09-30 | 北京工业大学 | Method for dynamic allocation of Map/Reduce data processing platform memory resources based on prediction |
CN106027318A (en) * | 2016-07-24 | 2016-10-12 | 成都育芽科技有限公司 | Cloud computing-based two-level optimal scheduling management platform for virtual machine |
Non-Patent Citations (1)
Title |
---|
HIEN NGUYEN VAN,等: "SLA-aware Virtual Resource Management for Cloud Infrastructures", 《2009 NINTH IEEE INTERNATIONAL CONFERENCE ON COMPUTER AND INFORMATION TECHNOLOGY》 * |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107391257B (en) * | 2017-06-30 | 2020-10-13 | 北京奇虎科技有限公司 | Method, device and server for estimating memory capacity required by service |
CN107391257A (en) * | 2017-06-30 | 2017-11-24 | 北京奇虎科技有限公司 | Predictor method, device and the server of memory size needed for business |
CN107391627A (en) * | 2017-06-30 | 2017-11-24 | 北京奇虎科技有限公司 | EMS memory occupation analysis method, device and the server of data |
CN107391627B (en) * | 2017-06-30 | 2020-11-03 | 北京奇虎科技有限公司 | Data memory occupation analysis method and device and server |
CN107463642A (en) * | 2017-07-19 | 2017-12-12 | 北京京东尚科信息技术有限公司 | The method and apparatus for lifting Tool for Data Warehouse resource utilization |
CN110019110B (en) * | 2017-07-28 | 2022-11-18 | 腾讯科技(深圳)有限公司 | Capacity management method, device and equipment of service system and service system |
CN110019110A (en) * | 2017-07-28 | 2019-07-16 | 腾讯科技(深圳)有限公司 | A kind of capacity management methods of operation system, device, equipment and operation system |
CN109561054A (en) * | 2017-09-26 | 2019-04-02 | 华为技术有限公司 | A kind of data transmission method, controller and access device |
CN108595367A (en) * | 2018-04-25 | 2018-09-28 | 银川华联达科技有限公司 | A kind of server system based on computer cluster in LAN |
CN108595367B (en) * | 2018-04-25 | 2021-12-10 | 广州高专资讯科技有限公司 | Server system based on computer cluster in local area network |
CN109165093B (en) * | 2018-07-31 | 2022-07-19 | 宁波积幂信息科技有限公司 | System and method for flexibly distributing computing node cluster |
CN109165093A (en) * | 2018-07-31 | 2019-01-08 | 宁波积幂信息科技有限公司 | A kind of calculate node cluster elasticity distribution system and method |
CN109271251A (en) * | 2018-08-09 | 2019-01-25 | 深圳市瑞云科技有限公司 | A method of by constraining come scheduling node machine |
CN112685170A (en) * | 2019-10-18 | 2021-04-20 | 伊姆西Ip控股有限责任公司 | Dynamic optimization of backup strategies |
CN112685170B (en) * | 2019-10-18 | 2023-12-08 | 伊姆西Ip控股有限责任公司 | Dynamic optimization of backup strategies |
CN111552550A (en) * | 2020-04-26 | 2020-08-18 | 星环信息科技(上海)有限公司 | Task scheduling method, device and medium based on GPU (graphics processing Unit) resources |
CN111679911A (en) * | 2020-06-04 | 2020-09-18 | 中国建设银行股份有限公司 | Management method, device, equipment and medium for GPU (graphics processing Unit) card in cloud environment |
CN111679911B (en) * | 2020-06-04 | 2024-01-16 | 建信金融科技有限责任公司 | Management method, device, equipment and medium of GPU card in cloud environment |
CN113806011A (en) * | 2021-08-17 | 2021-12-17 | 曙光信息产业股份有限公司 | Cluster resource control method and device, cluster and computer readable storage medium |
CN113806011B (en) * | 2021-08-17 | 2023-12-19 | 曙光信息产业股份有限公司 | Cluster resource control method and device, cluster and computer readable storage medium |
CN116112497A (en) * | 2022-12-29 | 2023-05-12 | 天翼云科技有限公司 | Node scheduling method, device, equipment and medium of cloud host cluster |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106790636A (en) | A kind of equally loaded system and method for cloud computing server cluster | |
US9015227B2 (en) | Distributed data processing system | |
US7627618B2 (en) | System for managing data collection processes | |
CN108845874B (en) | Dynamic resource allocation method and server | |
CN109062658A (en) | Realize dispatching method, device, medium, equipment and the system of computing resource serviceization | |
CN107465708A (en) | A kind of CDN bandwidth scheduling systems and method | |
US20100131324A1 (en) | Systems and methods for service level backup using re-cloud network | |
Bhatia et al. | Htv dynamic load balancing algorithm for virtual machine instances in cloud | |
CN102662764B (en) | A kind of dynamic cloud computational resource optimizing distribution method based on SMDP | |
CN110071965B (en) | Data center management system based on cloud platform | |
TW201424305A (en) | CDN load balancing in the cloud | |
CN106817499A (en) | A kind of resources for traffic dispatching method and forecast dispatching device | |
CN109861850B (en) | SLA-based stateless cloud workflow load balancing scheduling method | |
CN106294472A (en) | The querying method of a kind of Hadoop data base HBase and device | |
CN106161485B (en) | A kind of resource regulating method of infrastructure service cluster, device and system | |
US20170339069A1 (en) | Allocating Cloud Computing Resources In A Cloud Computing Environment | |
CN107111520A (en) | Method and system for the real time resources consumption control in DCE | |
Nazar et al. | Modified shortest job first for load balancing in cloud-fog computing | |
CN107967175A (en) | A kind of resource scheduling system and method based on multiple-objection optimization | |
CN102932271A (en) | Method and device for realizing load balancing | |
Choi et al. | An improvement on the weighted least-connection scheduling algorithm for load balancing in web cluster systems | |
Petrovska et al. | Features of the distribution of computing resources in cloud systems | |
Hu et al. | Reducing access latency in erasure coded cloud storage with local block migration | |
CN111400033A (en) | Platform resource cost allocation method and device, storage medium and computer equipment | |
CN112749008A (en) | Cloud resource distribution system based on OpenStack and construction method thereof |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170531 |
|
RJ01 | Rejection of invention patent application after publication |