CN103279184A - Cloud server energy-saving system based on data mining - Google Patents
Cloud server energy-saving system based on data mining Download PDFInfo
- Publication number
- CN103279184A CN103279184A CN201310219589XA CN201310219589A CN103279184A CN 103279184 A CN103279184 A CN 103279184A CN 201310219589X A CN201310219589X A CN 201310219589XA CN 201310219589 A CN201310219589 A CN 201310219589A CN 103279184 A CN103279184 A CN 103279184A
- Authority
- CN
- China
- Prior art keywords
- energy
- cloud server
- monitor
- load
- data mining
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a cloud server energy-saving system based on data mining. The cloud server energy-saving system based on data mining comprises a monitoring system portion, an acquisition system portion, an automatic strategy-analyzing and strategy-implementing system portion connected with both the monitoring system portion and the acquisition system portion, and an energy saving master control system portion connected with the automatic strategy-analyzing and strategy-implementing system portion. The cloud server energy-saving system based on data mining has the advantages that energy saving control strategies suitable for current environmental states are automatically generated, complexity of cloud server architecture in the original system structure is broken through, operating modes of various devices in the system are automatically regulated and set, loads are configured among the devices in a centralized manner, and accordingly energy consumption of the cloud server system is effectively reduced while system stability is guaranteed and performances satisfactory to application requirements are provided.
Description
Technical field
The invention belongs to the computer communication technology field, relate to a kind of Cloud Server energy conserving system based on data mining.
Background technology
Cloud computing is the service supply model of a kind of based on network support isomery facility and resource circulation, and it offers the client can autonomous service, realizes the distribution according to need of resource, charges according to quantity.Cloud computing causes resource extentization, promotes the specialization of the division of labor, makes resource supplier and user all pay close attention to the business of oneself more, is conducive to reduce the unit resource cost, promotes the Network innovation.
Conventional needle is to take pre-definedly to Cloud Server method for managing system and strategy substantially, namely just decides in the data deployment, fully according to fixed mode and flow process running; Adjust these predefined management method and strategies if desired, just need the system manager that these methods and strategy are reconfigured.Because the complicacy of Cloud Server self framework, move the diversity of using on it in addition, complicated variation can take place in Cloud Server temperature, energy consumption and load etc., and fixed mode operating strategy is difficult to satisfy the lasting usability requirements of Cloud Server system under complex environment.If manually manage reconfiguring of method and strategy and rely on, along with the raising of Cloud Server system complexity and the frequent variations of running status thereof, the Cloud Server system manager will have to run around all the time wears him out, and probability of errors also will increase greatly.Simultaneously, for satisfying the demand of load peak, the data of current operation can keep the resource redundancy of significant proportion usually, and load is in relatively low level mostly in the reality.In this case, a large amount of hardware devices do not provide effective performance output, but energy consumption does not have reduction.Compare with traditional server system steady job pattern, the Cloud Server system emphasizes distribution according to need and resilient expansion, therefore traditional server system fixing configuration, power managed mode be can not adopt, must real-time, intelligent system power dissipation control, management be carried out according to application demand intensity.
Summary of the invention
For addressing the above problem, the object of the present invention is to provide a kind of Cloud Server energy conserving system based on data mining, to reduce the energy resource consumption of Cloud Server system effectively.
For achieving the above object, technical scheme of the present invention is:
A kind of Cloud Server energy conserving system based on data mining includes supervisory system part, acquisition system part, connects supervisory system part and the automatic strategy of acquisition system part respectively and derive with automatic tactful implementation system part and is connected automatically strategy and derive and automatic tactful implementation system energy-conservation master control system part partly; Wherein, described supervisory system part and acquisition system part are at the load information of keystone resources and the data of each keystone resources ADMINISTRATION SUBSYSTEM running status of operating system, non-linear, complication system operational process and feature are effectively analyzed, set up performance, load and energy consumption realistic model, and produce the energy-saving run strategy that is suitable for.
Further, described supervisory system partly includes operating system nucleus task counter, operating system nucleus scheduler program monitor, operating system nucleus load balancing monitor, operating system nucleus Memory Allocation labormonitor, operating system nucleus interruption route monitor, operating system nucleus cache monitor and operating system nucleus exchange area monitor and operating system nucleus network protocol stack monitor.
Further, described acquisition system partly includes processor load monitor, Installed System Memory load monitor, system's memory load monitor and grid load monitor.
Further, described energy-saving run strategy mainly is power, the operational mode of adjusting each hardware resource blocks, reduces the energy consumption of indivedual hardware resource blocks under the prerequisite that guarantees system stability and application performance demand as much as possible.
Further; described energy-saving run strategy mainly is under the prerequisite that guarantees system stability and application performance demand; the load of whole monitored system is allocated in more concentrated mode; thereby can be with part cloud computing device standby, shut down or be set to be equal to other states therewith, reduce with the energy consumption that realizes higher degree.
Further, processor management unit, memory management unit, network management unit and memory management unit in the described energy master control system part connected system, and effectively these resource control unit are adjusted and arranged flexibly, and configuration load between multiple devices in a concentrated manner.
Compared to prior art, a kind of Cloud Server system based on data mining of the present invention automatically generates and is suitable for current environment state Energy Saving Control strategy, break through in the original system structure because the complicacy of Cloud Server self framework, automatically adjust and arrange the operational mode of various kinds of equipment in the system, in a concentrated manner configuration load between multiple devices; Finally be implemented in and guarantee that system stability provides the performance that the satisfies application demand while, reduces the energy resource consumption of Cloud Server system effectively.
Description of drawings
Fig. 1 is the Organization Chart that the present invention is based on the Cloud Server system of data mining;
Fig. 2 is each component load of the present invention and power consumption operation principle of optimality diagram.
Embodiment
In order to make purpose of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explaining the present invention, and be not used in restriction the present invention.
As shown in Figure 1 and Figure 2, a kind of Cloud Server energy conserving system based on data mining of the present invention includes supervisory system part, acquisition system part, connects supervisory system part and the automatic strategy of acquisition system part respectively and derive with automatic tactful implementation system part and is connected automatically strategy and derive and automatic tactful implementation system energy-conservation master control system part partly.
Supervisory system part and acquisition system part are at the load information of keystone resources and the data of each keystone resources ADMINISTRATION SUBSYSTEM running status of operating system, on this basis non-linear, complication system operational process and feature are effectively analyzed, set up performance, load and energy consumption realistic model, and produce the energy-saving run strategy that is suitable for.According to described energy-saving run strategy, the operational mode of various kinds of equipment in the system, configuration load between multiple devices are in a concentrated manner automatically adjusted and arranged to the primary control program in the energy-conservation master control system part; Finally be implemented in and guarantee that system stability provides the performance that the satisfies application demand while, reduces the energy resource consumption of Cloud Server system effectively.
Described supervisory system partly includes operating system nucleus task counter, operating system nucleus scheduler program monitor, operating system nucleus load balancing monitor, operating system nucleus Memory Allocation labormonitor (not shown), operating system nucleus interruption route monitor, operating system nucleus cache monitor and operating system nucleus exchange area monitor (not shown) and operating system nucleus network protocol stack monitor (not shown); Described acquisition system partly includes processor load monitor, Installed System Memory load monitor, system's memory load monitor and grid load monitor.
Supervisory system part and acquisition system part are accurately, obtain the load information of system core resource and the running state data of each keystone resources ADMINISTRATION SUBSYSTEM of operating system in time, with the input as automatic tactful derivation system part, plan by artificial neural network to non-linear, effective identification and the prediction of complication system operational process and feature, employing is based on three layers of feedforward artificial neural network subsystem of BP algorithm, and the Disturbance Rejection system that will reduce enchancement factor effectively moves the shake of strategy as the necessary condition that stops neural metwork training, set up performance, the realistic model of load and energy consumption is suitable for current environment state Energy Saving Control strategy thereby automatically generate.
Described Energy Saving Control strategy is divided into two big classes: the one, adjust the power, operational mode of each hardware resource blocks etc., and under the prerequisite that guarantees system stability and application performance demand, reduce the energy consumption of indivedual hardware resource blocks as much as possible; The 2nd, under the prerequisite that guarantees system stability and application performance demand; the load of whole monitored system is allocated in more concentrated mode; thereby can be with part cloud computing device standby, shut down or be set to be equal to other states therewith, reduce with the energy consumption that realizes higher degree.
Set up energy-conservation master control system part, to realize parsing and the enforcement to the energy-saving run strategy of automatic tactful derivation system generation, energy-conservation master control system part is with processor management unit, memory management unit, network management unit and memory management unit etc. in the efficient way connected system, and effectively these resource control unit are adjusted and arranged flexibly, and configuration load between multiple devices in a concentrated manner.
Be an embodiment with the local node power managed strategy based on the Enhanced-conservative algorithm, processor is the establishment of consuming energy most in the system, the energy consumption that studies show that processor takies about 40% of total power consumption, and the energy consumption that reduces processor can obviously reduce total energy consumption.Relation between CPU power consumption and cpu frequency such as formula:
, wherein P is cpu power, and f is cpu frequency, and A and B are all constant, satisfy simultaneously
When CPU is comparatively idle, can reduce processor speed though reduce frequency, can so neither lose performance and reduce power consumption again by improving the cpu busy percentage compensating loss of energy.(Dynamic Voltage and Frequency Scaling DVFS) by adjusting the CPU electric voltage frequency, reduces the idle waste that produces of CPU to the dynamic voltage adjustment frequency modulation technology, thereby improves the efficiency ratio of node.General conservative modulator uses cpu busy percentage as the foundation of regulation voltage frequency, progressively improve frequency during greater than higher limit in utilization factor, progressively reduce frequency during less than lower limit in utilization factor, but it is bad that modulator all changes when unsmooth effect based on the Past algorithm in utilization factor, and when load reduces, frequency progressively descends, prolonged the time that frequency is adjusted, also increased the expense that frequency descends and brings, for improving this problem, employing is based on the Enhanced-conservative algorithm of conservative modulator, this algorithm uses the utilization factor of a plurality of timeslices as the foundation of adjusting frequency, and give different weights to the utilization factor of different time sheet, algorithm when utilization factor is lower than lower limit directly frequency be set to minimum frequency, though the decline of frequency may cause the rising of cpu busy percentage, can not surpass higher limit.
Fuzzy association rules digging technology with Cloud Server load and power consumption is that an embodiment describes, and the present invention proposes to determine method based on each component load of server and optimised power consumption desired value that fuzzy association rules is excavated.From each component load of server and power consumption actual operating data, utilize data mining technology that the historical data of reflection each component load of server and power consumption is analyzed, for the related personnel provides optimized operation mode and the parameter control of server under the application of different business kind and external condition, realize that the energy-conservation profound level of server of data-driven is processed with integrated.
Each component load of server and power consumption based on data mining are moved principle of optimality figure as shown in Figure 2, Data Warehouse is from database, model bank and knowledge base deposit senior application required or data dig algorithm model and knowledge according to gained, they can call mutually and share.Based on data warehouse, from historical data, extract the some operational factor optimal value models that obtain covering the object operation area through data mining, set up the moving model storehouse through regretional analysis, and form a cover by the inference rule of procedure parameter to the optimization model collection.According to the on-the-spot real time data that collects, knowledge and the regular parameter operation optimal objective value of determining under the current working of utilizing off-line to excavate are used for instructing the related personnel to adjust the corresponding component parameter and move to optimize.
At first be determining of data mining target, in each component load of server is analyzed, therefore actual loading and power consumption select it as server energy saving interpretational criteria the more comprehensive response service device system power dissipation situation of association analysis, and its related operational factor is optimized analysis.Load and power consumption are influenced by several factors relevance, mainly by the occupancy of CPU and internal memory, I/O and output power of power supply, rotation speed of the fan etc.This problem is excavated the historical data under the main adjustable parameters steady running condition that influences relevance, and to determine the optimal objective value of main operational factor, the result of optimization makes server system least in power-consuming.
Carry out data and select and the data pre-service, server running load and power consumption are to be in dynamic changing process, should select under the different service types data under the system stable operation operating mode to analyze.The data that collect are carried out validity checking, reject redundant attributes, clear data and misdata, all data are carried out translation and compress processing to eliminate the false judgment that dimension difference causes between the different qualities parameter, this programme adopts the extreme value standardization that data standard is turned to [0,1] interval, data-standardizing formula is as follows:
In the formula
After data,
It is respectively raw data
Bound, its data are automatically calculated by set, or specific area expert estimation obtains.
Improved fuzzy association rules mining algorithm, fuzzy association rules be shape as
Implications, wherein
,
, and
, namely X and Y are respectively two fogy project set, and they do not have common project.In general, correlation rule
All be to be choice criteria with minimum support and min confidence, its support s has described data item collection X and Y and has appeared at probability in the same affairs simultaneously, and degree of confidence c has referred to occur in the affairs of data item collection X, the probability that data item collection Y also occurs.They can be expressed as respectively
Wherein:
Subordinate function for X; | D| represents the affairs sum of data centralization.The purpose of data mining is to find out those credible and representational rules, minimum support
And min confidence
Specified the threshold values of support and degree of confidence, their separate provision correlation rule set up support and the degree of confidence of the minimum that must reach, namely
By introducing the correlation rule criterion; Interest-degree has guaranteed the validity of data mining results.
Minimum support and min confidence can not guarantee that usually the correlation rule of excavating all is user's interest, so this problem is introduced the concept of interest-degree to reduce insignificant rule when excavating with fuzzy association rules.Interest-degree is to characterize the user to the tolerance of the degree of concern of rule, and it is the user to the taking all factors into consideration of knowledge novelty, availability and the interpretation of excavating, and interest-degree numerical value is more big, and is more high to the degree of concern of rule.Its expression formula is
Utilize improved fuzzy association rules mining algorithm to produce the rule of expectation, the fuzzy association rules mining algorithm at first is converted into the fuzzy variable value of representing with subordinate function to each attribute, calculate the weights of each attribute corresponding fuzzy set in the transaction database then, utilize the algorithm (Apriori algorithm) of association rule mining to find out support greater than the sport collection of the given minimum support of user, these sport collection just produce the interested rule of people after treatment.Note to appear at candidate simultaneously except two subregions of same attribute
Same China and foreign countries, by
Generate
Process and the algorithm of association rule mining similar.
The above only is preferred embodiment of the present invention, not in order to limiting the present invention, all any modifications of doing within the spirit and principles in the present invention, is equal to and replaces and improvement etc., all should be included within protection scope of the present invention.
Claims (6)
1. Cloud Server energy conserving system based on data mining is characterized in that: include supervisory system part, acquisition system part, connect supervisory system part and the automatic strategy of acquisition system part respectively and derive with automatic tactful implementation system part and is connected automatically strategy and derive and automatic tactful implementation system energy-conservation master control system part partly; Wherein, described supervisory system part and acquisition system part are at the load information of keystone resources and the data of each keystone resources ADMINISTRATION SUBSYSTEM running status of operating system, non-linear, complication system operational process and feature are effectively analyzed, set up performance, load and energy consumption realistic model, and produce the energy-saving run strategy that is suitable for.
2. according to the described Cloud Server energy conserving system based on data mining of claim 1, it is characterized in that: described supervisory system partly includes operating system nucleus task counter, operating system nucleus scheduler program monitor, operating system nucleus load balancing monitor, operating system nucleus Memory Allocation labormonitor, operating system nucleus interruption route monitor, operating system nucleus cache monitor and operating system nucleus exchange area monitor and operating system nucleus network protocol stack monitor.
3. according to the described Cloud Server energy conserving system based on data mining of claim 2, it is characterized in that: described acquisition system partly includes processor load monitor, Installed System Memory load monitor, system's memory load monitor and grid load monitor.
4. according to the described Cloud Server energy conserving system based on data mining of claim 3, it is characterized in that: described energy-saving run strategy mainly is power, the operational mode of adjusting each hardware resource blocks, reduces the energy consumption of indivedual hardware resource blocks under the prerequisite that guarantees system stability and application performance demand as much as possible.
5. according to the described Cloud Server energy conserving system based on data mining of claim 3; it is characterized in that: described energy-saving run strategy mainly is under the prerequisite that guarantees system stability and application performance demand; the load of whole monitored system is allocated in more concentrated mode; thereby can be with part cloud computing device standby, shut down or be set to be equal to other states therewith, reduce with the energy consumption that realizes higher degree.
6. according to claim 4 or 5 described Cloud Server energy conserving systems based on data mining, it is characterized in that: processor management unit, memory management unit, network management unit and memory management unit in the described energy master control system part connected system, and effectively these resource control unit are adjusted and arranged flexibly, and configuration load between multiple devices in a concentrated manner.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310219589XA CN103279184A (en) | 2013-06-05 | 2013-06-05 | Cloud server energy-saving system based on data mining |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310219589XA CN103279184A (en) | 2013-06-05 | 2013-06-05 | Cloud server energy-saving system based on data mining |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103279184A true CN103279184A (en) | 2013-09-04 |
Family
ID=49061743
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310219589XA Pending CN103279184A (en) | 2013-06-05 | 2013-06-05 | Cloud server energy-saving system based on data mining |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103279184A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103902016A (en) * | 2014-04-28 | 2014-07-02 | 浪潮电子信息产业股份有限公司 | Server power consumption management method oriented to scene prediction |
| CN104298536A (en) * | 2014-10-09 | 2015-01-21 | 南京大学镇江高新技术研究院 | Dynamic frequency modulation and pressure adjustment technology based data center energy-saving dispatching method |
| CN104391560A (en) * | 2014-09-28 | 2015-03-04 | 浪潮(北京)电子信息产业有限公司 | Hopfield neural network-based server energy-saving method and device for cloud data center |
| CN104503843A (en) * | 2014-12-25 | 2015-04-08 | 浪潮电子信息产业股份有限公司 | Power consumption managing method and device |
| CN111522255A (en) * | 2020-04-22 | 2020-08-11 | 第四范式（北京）技术有限公司 | Simulation system and simulation method |
| CN117421186A (en) * | 2023-12-18 | 2024-01-19 | 苏州元脑智能科技有限公司 | Method, device, system and storage medium for adjusting server operation parameters |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102419589A (en) * | 2011-12-31 | 2012-04-18 | 国网信息通信有限公司 | Intelligent power utilization system and method for district |
| CN102541015A (en) * | 2012-01-10 | 2012-07-04 | 深圳市嘉力达实业有限公司 | Intelligent energy efficiency control system and method |
| CN102567396A (en) * | 2010-12-30 | 2012-07-11 | 中国移动通信集团公司 | Method, system and device for data mining on basis of cloud computing |
| CN102937799A (en) * | 2012-09-10 | 2013-02-20 | 山东省计算中心 | Data center energy saving system and method |
| CN103078955A (en) * | 2013-01-29 | 2013-05-01 | 东北大学 | Device and method for transferring state of cloud computing platform server |
-
2013
- 2013-06-05 CN CN201310219589XA patent/CN103279184A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102567396A (en) * | 2010-12-30 | 2012-07-11 | 中国移动通信集团公司 | Method, system and device for data mining on basis of cloud computing |
| CN102419589A (en) * | 2011-12-31 | 2012-04-18 | 国网信息通信有限公司 | Intelligent power utilization system and method for district |
| CN102541015A (en) * | 2012-01-10 | 2012-07-04 | 深圳市嘉力达实业有限公司 | Intelligent energy efficiency control system and method |
| CN102937799A (en) * | 2012-09-10 | 2013-02-20 | 山东省计算中心 | Data center energy saving system and method |
| CN103078955A (en) * | 2013-01-29 | 2013-05-01 | 东北大学 | Device and method for transferring state of cloud computing platform server |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103902016A (en) * | 2014-04-28 | 2014-07-02 | 浪潮电子信息产业股份有限公司 | Server power consumption management method oriented to scene prediction |
| CN104391560A (en) * | 2014-09-28 | 2015-03-04 | 浪潮(北京)电子信息产业有限公司 | Hopfield neural network-based server energy-saving method and device for cloud data center |
| CN104298536A (en) * | 2014-10-09 | 2015-01-21 | 南京大学镇江高新技术研究院 | Dynamic frequency modulation and pressure adjustment technology based data center energy-saving dispatching method |
| CN104503843A (en) * | 2014-12-25 | 2015-04-08 | 浪潮电子信息产业股份有限公司 | Power consumption managing method and device |
| CN111522255A (en) * | 2020-04-22 | 2020-08-11 | 第四范式（北京）技术有限公司 | Simulation system and simulation method |
| CN117421186A (en) * | 2023-12-18 | 2024-01-19 | 苏州元脑智能科技有限公司 | Method, device, system and storage medium for adjusting server operation parameters |
| CN117421186B (en) * | 2023-12-18 | 2024-03-15 | 苏州元脑智能科技有限公司 | Method, device, system and storage medium for adjusting server operation parameters |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103279184A (en) | Cloud server energy-saving system based on data mining | |
| CN102004671B (en) | Resource management method of data center based on statistic model in cloud computing environment | |
| CN103384272B (en) | A kind of cloud service distributive data center system and load dispatching method thereof | |
| Zhang et al. | Electricity bill capping for cloud-scale data centers that impact the power markets | |
| Liu et al. | Restoration of power distribution systems with multiple data centers as critical loads | |
| CN105022823B (en) | A kind of cloud service performance early warning event generation method based on data mining | |
| US10054997B2 (en) | Multivariable control for power-latency management to support optimization of data centers or other systems | |
| US10048732B2 (en) | Datacenter power management system | |
| CN116881085A (en) | Method for optimizing energy consumption of server | |
| Li et al. | Load prediction-based automatic scaling cloud computing | |
| US20180006964A1 (en) | Datacenter byproduct management interface system | |
| CN116843152A (en) | Electric power-data service-oriented Internet data center double-layer planning method | |
| CN107169562A (en) | A kind of energy efficiency controlling method and device for data center | |
| CN117096930A (en) | Application method and system of distributed resource group modulation group control technology | |
| US20180006966A1 (en) | Infrastructure resource management system | |
| CN110531680B (en) | Intelligent power monitoring deviation power supply system and control method | |
| Alimisis et al. | Operation and control zones for future complex power systems | |
| CN113268130A (en) | Server power supply method, storage medium and computer equipment | |
| Daorui et al. | Optimized Energy Distribution in Smart Grid System Using Hybrid Machine Learning Techniques | |
| US11962152B2 (en) | Method for supervisory control of building power consumption | |
| Liang et al. | A Balanced Scheduling Algorithm for Virtual Machines in Cloud Data Centers Based on Dynamic Machine Learning Algorithm | |
| CN118505437B (en) | Distributed micro measurement and control system based on comprehensive energy system | |
| Wan et al. | Time series prediction method of metrics of dispatching automation system based on ai platform | |
| US20240104478A1 (en) | Data center management system | |
| Lu | Research on Elastic Scheduling of Private Cloud Resources in Mobile Social Networks Based on Improved Particle Swarm Optimization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130904 |