CN104423531A - Data center energy consumption scheduling method and data center energy consumption scheduling device - Google Patents
Data center energy consumption scheduling method and data center energy consumption scheduling device Download PDFInfo
- Publication number
- CN104423531A CN104423531A CN201310399815.7A CN201310399815A CN104423531A CN 104423531 A CN104423531 A CN 104423531A CN 201310399815 A CN201310399815 A CN 201310399815A CN 104423531 A CN104423531 A CN 104423531A
- Authority
- CN
- China
- Prior art keywords
- rack
- data center
- resource utilization
- environment temperature
- load
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/26—Power supply means, e.g. regulation thereof
- G06F1/32—Means for saving power
- G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
- G06F1/3206—Monitoring of events, devices or parameters that trigger a change in power modality
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F1/00—Details not covered by groups G06F3/00 - G06F13/00 and G06F21/00
- G06F1/26—Power supply means, e.g. regulation thereof
- G06F1/32—Means for saving power
- G06F1/3203—Power management, i.e. event-based initiation of a power-saving mode
- G06F1/3234—Power saving characterised by the action undertaken
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention provides a data center energy consumption scheduling method and a data center energy consumption scheduling device. The method comprises the following steps of acquiring resource utilization rate and/or environment temperature of one or a plurality of cabinets of a data center; and scheduling energy consumption of the data center according to the acquired resource utilization rate and/or the environment temperature. By the data center energy consumption scheduling method and the data center energy consumption scheduling device, the problems that reduction treatment on the energy consumption of a data center is incomplete by using a correlation technique and the energy using efficiency of the data center is not high are solved, energy consumption scheduling of load distribution can be realized, and energy consumption of refrigeration equipment of the data center is reduced to a certain degree.
Description
Technical field
The present invention relates to the communications field, in particular to a kind of consumption of data center scheduling processing method and device.
Background technology
Along with enterprise is to the growth of the information processing capability demand that cloud computing provides, can imagine that data center will become indispensable public infrastructure as generating plant.Data center's quantity that this trend is built from setting up the project one after another in various places just can be verified, and newly-built data center's scale is expanded rapidly simultaneously, and bulk density and energy requirement (comprising refrigeration facility and computing equipment) also increase fast.As can be seen from the operation data of existing data center; the spending of energy supply becomes one of main expenditure in data center's operation process gradually; how to reduce unnecessary energy consumption; improve the energy utilization efficiency of existing equipment; reduce the energy spending of whole data center, and then reduce greenhouse gas emission protection of the environment.
In the related, to the research of data central energy service efficiency, mainly concentrate on and how to reduce calculating energy consumption by Intel Virtualization Technology.Based on the characteristic of virtual machine in the integration of computational resource and the migration of calculation task etc., Internet data center (Internet Data Center, referred to as IDC) predict that the enterprise more than 50% in recent years will be had for the first time to apply to be operated in virtual machine, simultaneously annual will have the server that dispatches from the factory more than 23% to support Intel Virtualization Technology, and namely these servers dispatched from the factory will pre-installation virtual machine monitor.The important foundation that Intel Virtualization Technology will become the application of following enterprise can be predicted.For data center's administration of energy conservation, Intel Virtualization Technology is also important ingredient.
The research of data central energy service efficiency is being concentrated on to how reduced by Intel Virtualization Technology that calculate can be consuming time, load balancing between Main Basis server, but the heat distribution affecting data center also relates to many factors, there is not the technical finesse of the heat distribution of data center being unified to consideration in correlation technique.
Therefore, exist in the related and process is reduced comprehensively to the energy consumption of data center, cause the problem that consumption of data center utilization ratio is not high.
Summary of the invention
The invention provides a kind of consumption of data center scheduling processing method and device, at least to solve to exist in the related, process is reduced comprehensively to the energy consumption of data center, cause the problem that consumption of data center utilization ratio is not high.
According to an aspect of the present invention, provide a kind of consumption of data center scheduling processing method, comprising: the resource utilization and/or the environment temperature that obtain the one or more rack of data center; According to the described resource utilization obtained and/or environment temperature, described consumption of data center is dispatched.
Preferably, scheduling is carried out to described consumption of data center comprise according to the described environment temperature obtained: judge in the described one or more rack obtained in predetermined time section, whether the highest environment temperature has reduction; When judged result is for being, improve the feed air temperature of described data center refrigeration plant.
Preferably, carry out scheduling according to the described resource utilization obtained and described environment temperature to described consumption of data center to comprise: distribute rack according to the described resource utilization obtained and described environment temperature determination load; Distribute rack according to the described load determined to dispatch described consumption of data center.
Preferably, determine that described load distributes rack and comprises according to the described resource utilization obtained and described environment temperature: determine that the resource utilization of the described one or more rack obtained and described environment temperature are sample value; According to described sample value, predict resource utilization predicted value and the environment temperature predicted value of described one or more rack after described one or more rack increases load newly; Determine when described resource utilization predicted value nonoverload, the rack of the minimum correspondence of described environment temperature predicted value is that load distributes rack.
Preferably, determining when described resource utilization predicted value nonoverload, the rack of the minimum correspondence of described environment temperature predicted value is, after described load distributes rack, also comprise: add up the error between the resource utilization of the actual described one or more rack obtained after newly-increased load and ambient temperature value and described prediction resource utilization and described environment temperature predicted value; According to sample value described in described error update.
According to a further aspect in the invention, provide a kind of consumption of data center dispatch deal device, comprising: acquisition module, for obtaining resource utilization and/or the environment temperature of the one or more rack of data center; Scheduler module, for dispatching described consumption of data center according to the described resource utilization obtained and/or environment temperature.
Preferably, described scheduler module comprises: judging unit, for judging in described one or more rack of obtaining in predetermined time section, whether the highest environment temperature has reduction; Improving unit, for when the judged result of above-mentioned judging unit is for being, improving the feed air temperature of described data center refrigeration plant.
Preferably, described scheduler module comprises: determining unit, for distributing rack according to the described resource utilization obtained and described environment temperature determination load; Scheduling unit, dispatches described consumption of data center for distributing rack according to the described load determined.
Preferably, described determining unit comprises: first determines subelement, for determining that resource utilization and the described environment temperature of the described one or more rack obtained are sample value; Predictor unit, for according to described sample value, predicts resource utilization predicted value and the environment temperature predicted value of described one or more rack after described one or more rack increases load newly; Second determines subelement, and for determining when described resource utilization predicted value nonoverload, the rack of the minimum correspondence of described environment temperature predicted value is that load distributes rack.
Preferably, described determining unit also comprises: statistics subelement, for adding up after newly-increased load the error between the resource utilization of the actual described one or more rack obtained and ambient temperature value and described prediction resource utilization and described environment temperature predicted value; Upgrade subelement, for according to sample value described in described error update.
By the present invention, adopt the resource utilization and/or environment temperature that obtain the one or more rack of data center; According to the described resource utilization obtained and/or environment temperature, described consumption of data center is dispatched, solve in correlation technique the energy consumption minimizing process existed data center not comprehensive, cause the problem that consumption of data center utilization ratio is not high, and then reach the energy consumption scheduling that can not only realize load and distribute, and advantageously reduce the effect of data center's refrigeration plant energy consumption to a certain extent.
Accompanying drawing explanation
Accompanying drawing described herein is used to provide a further understanding of the present invention, and form a application's part, schematic description and description of the present invention, for explaining the present invention, does not form inappropriate limitation of the present invention.In the accompanying drawings:
Fig. 1 is the process flow diagram of the consumption of data center scheduling processing method according to the embodiment of the present invention;
Fig. 2 is the structured flowchart of the consumption of data center dispatch deal device according to the embodiment of the present invention;
Fig. 3 is the preferred structure block diagram one of scheduler module 24 in the consumption of data center dispatch deal device according to the embodiment of the present invention;
Fig. 4 is the preferred structure block diagram two of scheduler module 24 in the consumption of data center dispatch deal device according to the embodiment of the present invention;
Fig. 5 is the structured flowchart of determining unit 42 in scheduler module 24 in the consumption of data center dispatch deal device according to the embodiment of the present invention;
Fig. 6 is the preferred structure block diagram of determining unit 42 in scheduler module 24 in the consumption of data center dispatch deal device according to the embodiment of the present invention;
Fig. 7 is the view of each rack in the frame according to the embodiment of the present invention.
Embodiment
Hereinafter also describe the present invention in detail with reference to accompanying drawing in conjunction with the embodiments.It should be noted that, when not conflicting, the embodiment in the application and the feature in embodiment can combine mutually.
Provide a kind of consumption of data center scheduling processing method in the present embodiment, Fig. 1 is the process flow diagram of the consumption of data center scheduling processing method according to the embodiment of the present invention, and as shown in Figure 1, this flow process comprises the steps:
Step S102, obtain resource utilization and/or the environment temperature of the one or more rack of data center, wherein, this ambient humidity can be the temperature in of this one or more rack;
Step S104, dispatches consumption of data center according to the resource utilization obtained and/or environment temperature.
Pass through above-mentioned steps, not only load balancing between servers factor is considered by utilizing consumption of data center, and consider the environment temperature factor of whole data center, unilateral influence factor is only related to relative in correlation technique, not only solve in correlation technique the energy consumption minimizing process existed data center not comprehensive, cause the problem that consumption of data center utilization ratio is not high, and then reach the energy consumption scheduling that can not only realize load and distribute, and advantageously reduce the effect of data center's refrigeration plant energy consumption to a certain extent.
When consumption of data center being dispatched according to the environment temperature (being described for the temperature in of rack) obtained, judge in the one or more racks obtained in predetermined time section, whether the highest environment temperature has reduction; When judged result is for being, improving the feed air temperature of data center's refrigeration plant, namely by improving the feed air temperature of refrigeration plant, thus effectively reducing the energy consumption of refrigeration plant, reducing the energy consumption expense of data center to a certain extent.
When consumption of data center being dispatched according to the resource utilization obtained and environment temperature, also comprise and allocation process is carried out to the load of each rack, such as, rack (rack that namely newly-increased load is corresponding) can be distributed according to the resource utilization obtained and environment temperature determination load; Distribute rack according to this load determined to dispatch data power consumption, namely making the load increased newly distribute rack is distribute the most rational rack of load (rack that namely load is minimum), be unlikely to a newly-increased load to be assigned in the rack that will transship, and the few rack of load is in the state of the wasting of resources.
More preferably, distribute rack according to the resource utilization obtained and environment temperature determination load and can adopt following process, first determine that the resource utilization of one or more racks that obtains and environment temperature are sample value, namely record the current state of rack, and this current state is carried out the sample of load distribution as prediction module; Based on this sample value, the resource utilization predicted value of prediction one or more rack after one or more rack increases load newly and environment temperature predicted value, namely predict when newly-increased load is assigned on each rack, on the impact that the distribution of new heat produces, namely obtain the new state that each rack is possible; Determine when resource utilization predicted value nonoverload, the rack of the minimum correspondence of environment temperature predicted value is that load distributes rack, namely selects that rack of the minimum correspondence of the highest temperature under new state to be that optimum load distributes rack.
In order to improve the degree of accuracy of prediction, determining when resource utilization predicted value nonoverload, the rack of the minimum correspondence of environment temperature predicted value is after load distributes rack, also need to collect real data feedback simultaneously, namely after obtaining the load distribution rack newly-increased load being assigned to above-mentioned optimum, the resource utilization of each rack actual and environment temperature, then, according to the error between the resource utilization of the actual data statistics got actual one or more racks obtained after newly-increased load and ambient temperature value and prediction resource utilization and environment temperature predicted value, again according to the above-mentioned sample value for prognosis modelling decision-making of error update.
Additionally provide a kind of consumption of data center dispatch deal device in the present embodiment, this device is used for realizing above-described embodiment and preferred implementation, has carried out repeating no more of explanation.As used below, term " module " can realize the software of predetermined function and/or the combination of hardware.Although the device described by following examples preferably realizes with software, hardware, or the realization of the combination of software and hardware also may and conceived.
Fig. 2 is the structured flowchart of the consumption of data center dispatch deal device according to the embodiment of the present invention, and as shown in Figure 2, this device comprises acquisition module 22 and scheduler module 24, is described below to this device.
Acquisition module 22, for obtaining resource utilization and/or the environment temperature of the one or more rack of data center; Scheduler module 24, is connected to above-mentioned acquisition module 22, for dispatching consumption of data center according to the resource utilization obtained and/or environment temperature.
Fig. 3 is the preferred structure block diagram one of scheduler module 24 in the consumption of data center dispatch deal device according to the embodiment of the present invention, and as shown in Figure 3, this scheduler module 24 comprises judging unit 32 and improves unit 34, is described below to this scheduler module 24.
Judging unit 32, for judging in one or more racks of obtaining in predetermined time section, whether the highest environment temperature has reduction; Improving unit 34, be connected to above-mentioned judging unit 32, for when the judged result of above-mentioned judging unit 32 is for being, improving the feed air temperature of data center's refrigeration plant.
Fig. 4 is the preferred structure block diagram two of scheduler module 24 in the consumption of data center dispatch deal device according to the embodiment of the present invention, and as shown in Figure 4, this scheduler module 24 comprises determining unit 42 and scheduling unit 44, is described below to this scheduler module 24.
Determining unit 42, for distributing rack according to the resource utilization obtained and environment temperature determination load; Scheduling unit 44, is connected to above-mentioned determining unit 42, dispatches data power consumption for distributing rack according to the load determined.
Fig. 5 is the structured flowchart of determining unit 42 in scheduler module 24 in the consumption of data center dispatch deal device according to the embodiment of the present invention, as shown in Figure 5, this determining unit 42 comprise first determine subelement 52, predictor unit 54 and second determines subelement 56, below this determining unit 42 is described.
First determines subelement 52, for determining that resource utilization and the environment temperature of the one or more racks obtained are sample value; Predictor unit 54, is connected to above-mentioned first and determines subelement 52, for according to above-mentioned sample value, predicts resource utilization predicted value and the environment temperature predicted value of one or more rack after one or more rack increases load newly; Second determines subelement 56, is connected to above-mentioned predictor unit 54, and for determining when resource utilization predicted value nonoverload, the rack of the minimum correspondence of environment temperature predicted value is that load distributes rack.
Fig. 6 is the preferred structure block diagram of determining unit 42 in scheduler module 24 in the consumption of data center dispatch deal device according to the embodiment of the present invention, as shown in Figure 6, this determining unit 42 is except comprising all modules shown in Fig. 5, also comprise statistics subelement 62 and upgrade subelement 64, below this determining unit 42 being described.
Statistics subelement 62, is connected to above-mentioned second and determines subelement 56, for adding up after newly-increased load the error between the resource utilization of the actual one or more racks obtained and ambient temperature value and prediction resource utilization and environment temperature predicted value; Upgrade subelement 64, be connected to above-mentioned statistics subelement 62, for according to the above-mentioned sample value of above-mentioned error update.
Based in correlation technique, the distribution of data center's rack is cold, the layout type that hot corridor is alternate, the cold wind that refrigeration plant provides blows up (all can have between the floor at general data center and ground one deck for the circulation of cold air) every sky from the underground in " cold corridor " by seamed floor, enter from the front shroud entrance of rack, discharge from rack rear after mixing with the hot-air in rack, take away corresponding heat, therefore the corridor at the rear of rack is exactly " hot corridor ", these " hot blast " major parts all can siphon away from top by cooled equipment, but it is elegant to " cold corridor " also to have small part, thus affect the temperature in of the workstation in the middle of rack.Generally speaking, the temperature of workstation should be less than certain temperature specified by equipment supplier, just at last at the working environment of safety, such as, can be 32 degrees Celsius to used this temperature of small-sized data center, different station device has different safe working temperatures.The working load that the temperature in of workstation is also current with it is simultaneously relevant, and the larger heat that it produces of working load is also more, and the heat that " cold wind " can be taken away is certain; The difference of workstation manufacture material also has different conductivity properties in addition.
Therefore, each frame of data center is so that the temperature in of each rack all can be different, and cloud computing resources dispatching management information system in the past only considers the problem of load balancing between each server, or the angle only integrated from virtual machine reduces the energy consumption expense of cloud data center, and have ignored the waste of the unbalanced energy consumption for cooling brought of data center's heat distribution.In the present embodiment by the intelligent scheduling to data center-point load, the heat of data center is distributed and reaches balanced, thus the feed air temperature of refrigeration plant can be heightened, and then reduce the expense of energy consumption for cooling.Because at present in order to ensure the safe operation of data center's all devices, the feed air temperature of refrigeration plant is all relevant to current data center the highest rack temperature in, therefore when the highest rack temperature in of data center reduces, raising feed air temperature that just can be suitable, and then the energy consumption reducing refrigeration plant.
For in correlation technique, heat skewness weighing apparatus in data center and the problem that causes energy consumption for cooling excessive, provide a kind of load dispatching method based on heat perception in the present embodiment, the heat of data center is distributed and reaches balanced as far as possible, thus make the feed air temperature of refrigeration plant be able to suitable raising, and then reduce the expense of energy consumption for cooling.
Based on the cloud data center power-economizing method of heat perception, should comprise the following steps:
Step S1, the physical environment monitoring of data center: such as, the distribution of heat, refrigeration plant ruuning situation.The temperature in of each workstation in each frame entrance and rack in the heart in Monitoring Data.If Monitoring Data display indoor temperature is elevated to a certain predetermined alarm value, then automatically sends prompting and warning information to related personnel by network, allow it take measures in time to avoid computing equipment work long hours in high temperature environments and affect tenure of use; And whether supvr is normally run equipment by the service data of refrigeration plant and judges fast.In addition, also these information collected are carried out network centralized stores, and provide open interface for these data;
Step S2, data center's layer is based on the load dispatch strategy of heat perception: based on the model of neural network algorithm, by to the study of historical data and modeling, find the relation of each frame total load and its temperature in, and then distributed by the heat that the workstation that model prediction is newly assigned in different frame to load will bring, by comparing each heat distribution results, choosing optimum heat distribution results and deciding newly to arrive the workstation which frame load is assigned to;
Step S3, blade rack layer is based on the load dispatch strategy of heat perception: be also based on neural network and the model strengthening learning algorithm, by to the study of historical data and modeling, find the relation of each rack load and its temperature in frame, and then can predict that being newly assigned to load the heat that in frame, different rack will bring distributes by this model, by the comparison to each heat distribution results, choose optimum heat distribution results and decide newly to arrive the workstation which rack load is assigned to.
In order to monitor the physical environment of data center in step S1, such as, the distribution of heat and refrigeration plant ruuning situation etc., deploy temperature sensor before each frame and each rack, can the temperature in of each workstation in each frame of perception and rack; Also sensor is placed with to measure temperature, the rotating speed of air-conditioning inside and the setting etc. of safe temperature of air-conditioning supply cold air in the refrigeration plant of the heart in the data in addition.If indoor temperature is elevated to a certain predetermined alarm value, automatically can sends prompting and warning information to related personnel by network, allow it take measures in time to avoid computing equipment work long hours in high temperature environments and affect tenure of use; The service data of refrigeration plant can allow supvr judge fast, and whether this equipment normally runs.In addition, also these information collected are carried out network centralized stores, and provide open interface for these data.
The model prediction algorithm based on neural network algorithm related in step S2 and step S3 can comprise the steps:
First, need to collect certain sample data, comprise each rack resource utilization at set intervals, and the data such as the temperature in of this rack, use <U
i(t), T
it () > represents the state of each rack.Based on the basis of neural network prediction model, can fast by predicting the state-behavior of current environment,
（1）
In above formula (1), can see that neural network model is to new prediction mode of distributing behavior to load, the arrow left side is the state of system environments at time t, U
it () represents the resource utilization of workstation i in t, T
it () represents the temperature in of workstation i in t.The predicted state of system after being newly assigned to workstation i to load on the right of arrow, wherein T
i(t+1) ' be by U
i(t+1) ' calculated by neural network prediction model, U
i(t+1) ' can think when the t+1 period starts and be equal to U
it (), upgrades according to actual observed value at the end of t+1 period again.There are these to predict the outcome can to find easily temperature in mxm. wherein, distributed by more various load the result that behavior produces and find, so just can determine which workstation is most suitable load-receipt node.The load migration behavior of temperature in being crossed to high node also can find the most suitable receiving node of load by same method.After carry out behaviour decision making and execution according to predicting the outcome, also need to collect real system feedback simultaneously, i.e. real system environments temperature in mxm., the value collected is added in sample space, upgrade neural network prediction model according to new sample space at set intervals, improve the precision of prediction.
By above-described embodiment and preferred implementation, not only achieve the formulation of the intelligent load allocation strategy based on rack resource utilization and temperature in; And achieve a neural network model and combine with closed-loop control theory, the precision of prediction of heat distribution is improved by constantly updating neural network prediction model.Make the heat distribution of data center balanced as far as possible, and then reduce the expense of energy consumption for cooling, thus reach energy-conservation object.
Below in conjunction with accompanying drawing, the preferred embodiment for the present invention is described.
Should comprise the following steps based on the cloud data center power-economizing method of heat perception:
Step S1, the resource utilization of each rack and the monitoring of temperature in: passage is deployed on the sensor of data center and runs on the monitoring modular on each server and carry out Real-Time Monitoring to the resource utilization of each rack and temperature in, at set intervals, calculate and record once current rack resource utilization and inlet temperature <U
i(t), T
it () >, after all these information records complete, namely sends to neural network prediction module to carry out the formulation of the decision-making of load distribution;
Step S2, the formulation of the intelligent load allocation strategy of heat perception: neural network prediction module is according to the current state of each rack, prediction newly to load be assigned to after on each different rack, on the impact that the distribution of new heat produces, namely obtain the new state <U that each rack is possible
i(t+1), T
i(t+1) >, distributing difference selects predicting the outcome of generation to compare analysis, select that selection that the highest temperature under optimum load distribution and new state is minimum, Fig. 7 is the view of each rack in the frame according to the embodiment of the present invention, and what namely select in the figure 7 is then rack 3;
Step S3, load distributes execution and performs rear status monitoring collects: the distribution performing load according to the result of decision of step S2, monitors the rack status information after collecting load distribution simultaneously.Error statistics is carried out, sample data when last follow-up neural network prediction model upgrades according to the result observed and actual prediction result.
Step S4, the feed air temperature of adjustment refrigeration plant: judge whether refrigeration plant can suitably adjust its feed air temperature according to the result that status monitoring is collected, if the highest temperature in of monitoring result display decreases, then suitably can improve the feed air temperature of refrigeration plant, so just can reduce the energy consumption expense of refrigeration plant, reach the object of saving whole consumption of data center.
Wherein, the rack resource utilization in step S1 can be mainly represented by the service condition of cpu resource;
Neural network prediction model in step S2 mainly can by carrying out analysis modeling, the relation between each rack resource utilization and rack temperature in found, i.e. <U to the sample space collected
1(t), U
2(t), U
3(t) ..., U
i(t) ... > and <T
1(t), T
2(t), T
3(t) ..., T
i(t) ... mapping relations between >, and then selection course in step S2 is as follows:
1. first quantize newly arriving load, as needs how many cpu resources etc.This process need carries out resource requirement statistics to various dissimilar load.
2. predict that this load is assigned to each different rack i, <U
1(t), U
2(t), U
3(t) ... U
i(t)+△ u ... possible heat distribution <T corresponding to >
1(t+1), T
2(t+1), T
3(t+1) ..., T
i(t+1) ... >, and find maximum temperature in MaxT
i(t+1).
3. basis is assigned to the maximum temperature in { MaxT that different rack i finds
i}, and then find wherein minimum M in{MaxT (t+1)
i(t+1) i }, corresponding to minimum value is the rack finally will selected.
Used in step S3 is strengthen learning art online, by the precision of continuous learning improvement model.
The measure of the feed air temperature of the regulable control refrigeration plant in step S4 is main relevant to the interface that equipment supplier provides, and this control procedure can be completed by software control.
Obviously, those skilled in the art should be understood that, above-mentioned of the present invention each module or each step can realize with general calculation element, they can concentrate on single calculation element, or be distributed on network that multiple calculation element forms, alternatively, they can realize with the executable program code of calculation element, thus, they can be stored and be performed by calculation element in the storage device, and in some cases, step shown or described by can performing with the order be different from herein, or they are made into each integrated circuit modules respectively, or the multiple module in them or step are made into single integrated circuit module to realize.Like this, the present invention is not restricted to any specific hardware and software combination.
The foregoing is only the preferred embodiments of the present invention, be not limited to the present invention, for a person skilled in the art, the present invention can have various modifications and variations.Within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1. a consumption of data center scheduling processing method, is characterized in that, comprising:
Obtain resource utilization and/or the environment temperature of the one or more rack of data center;
According to the described resource utilization obtained and/or environment temperature, described consumption of data center is dispatched.
2. method according to claim 1, is characterized in that, carries out scheduling comprise according to the described environment temperature obtained to described consumption of data center:
Judge in the described one or more rack obtained in predetermined time section, whether the highest environment temperature has reduction;
When judged result is for being, improve the feed air temperature of described data center refrigeration plant.
3. method according to claim 1, is characterized in that, carries out scheduling comprise according to the described resource utilization obtained and described environment temperature to described consumption of data center:
Rack is distributed according to the described resource utilization obtained and described environment temperature determination load;
Distribute rack according to the described load determined to dispatch described consumption of data center.
4. method according to claim 1, is characterized in that, determines that described load distributes rack and comprises according to the described resource utilization obtained and described environment temperature:
Determine that the resource utilization of described one or more rack that obtains and described environment temperature are sample value;
According to described sample value, predict resource utilization predicted value and the environment temperature predicted value of described one or more rack after described one or more rack increases load newly;
Determine when described resource utilization predicted value nonoverload, the rack of the minimum correspondence of described environment temperature predicted value is that load distributes rack.
5. method according to claim 4, is characterized in that, is determining when described resource utilization predicted value nonoverload, and the rack of the minimum correspondence of described environment temperature predicted value is, after described load distributes rack, also comprise:
Error between the resource utilization of statistics actual described one or more rack obtained after newly-increased load and ambient temperature value and described prediction resource utilization and described environment temperature predicted value;
According to sample value described in described error update.
6. a consumption of data center dispatch deal device, is characterized in that, comprising:
Acquisition module, for obtaining resource utilization and/or the environment temperature of the one or more rack of data center;
Scheduler module, for dispatching described consumption of data center according to the described resource utilization obtained and/or environment temperature.
7. device according to claim 6, is characterized in that, described scheduler module comprises:
Judging unit, for judging in described one or more rack of obtaining in predetermined time section, whether the highest environment temperature has reduction;
Improving unit, for when the judged result of above-mentioned judging unit is for being, improving the feed air temperature of described data center refrigeration plant.
8. device according to claim 6, is characterized in that, described scheduler module comprises:
Determining unit, for distributing rack according to the described resource utilization obtained and described environment temperature determination load;
Scheduling unit, dispatches described consumption of data center for distributing rack according to the described load determined.
9. device according to claim 6, is characterized in that, described determining unit comprises:
First determines subelement, for determining that resource utilization and the described environment temperature of the described one or more rack obtained are sample value;
Predictor unit, for according to described sample value, predicts resource utilization predicted value and the environment temperature predicted value of described one or more rack after described one or more rack increases load newly;
Second determines subelement, and for determining when described resource utilization predicted value nonoverload, the rack of the minimum correspondence of described environment temperature predicted value is that load distributes rack.
10. device according to claim 9, is characterized in that, described determining unit also comprises:
Statistics subelement, for adding up after newly-increased load the error between the resource utilization of the actual described one or more rack obtained and ambient temperature value and described prediction resource utilization and described environment temperature predicted value;
Upgrade subelement, for according to sample value described in described error update.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310399815.7A CN104423531A (en) | 2013-09-05 | 2013-09-05 | Data center energy consumption scheduling method and data center energy consumption scheduling device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310399815.7A CN104423531A (en) | 2013-09-05 | 2013-09-05 | Data center energy consumption scheduling method and data center energy consumption scheduling device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104423531A true CN104423531A (en) | 2015-03-18 |
Family
ID=52972823
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310399815.7A Pending CN104423531A (en) | 2013-09-05 | 2013-09-05 | Data center energy consumption scheduling method and data center energy consumption scheduling device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104423531A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106096874A (en) * | 2016-08-18 | 2016-11-09 | 中国联合网络通信集团有限公司 | The energy efficiency managing method of data center, device and system |
| WO2016197849A1 (en) * | 2015-06-09 | 2016-12-15 | 邻元科技（北京）有限公司 | Control method, cooling apparatus system, cooling apparatus controller, cooling tower system, cooling tower controller, water pump system, and water pump controller |
| WO2018023496A1 (en) * | 2016-08-03 | 2018-02-08 | 深圳中兴力维技术有限公司 | Device racking management method and apparatus |
| CN108139942A (en) * | 2015-09-30 | 2018-06-08 | 华为技术有限公司 | The method of the end-to-end power efficiency modeling of data center |
| CN109189190A (en) * | 2018-10-16 | 2019-01-11 | 西安交通大学 | A kind of data center's thermal management method based on temperature prediction |
| CN109375994A (en) * | 2018-09-10 | 2019-02-22 | 西安交通大学 | Data center's task temperature prediction and dispatching method based on RBF neural |
| CN109800066A (en) * | 2018-12-13 | 2019-05-24 | 中国科学院信息工程研究所 | A kind of data center's energy-saving scheduling method and system |
| CN109890176A (en) * | 2019-03-01 | 2019-06-14 | 北京慧辰资道资讯股份有限公司 | A kind of method and device based on artificial intelligence optimization's energy consumption of machine room efficiency |
| CN110188932A (en) * | 2019-05-20 | 2019-08-30 | 国核电力规划设计研究院有限公司 | Consumption of data center prediction technique based on evaluation optimization |
| CN111258398A (en) * | 2020-01-10 | 2020-06-09 | 安擎(天津)计算机有限公司 | AI server fan speed adjusting method, system, terminal device and storage medium |
| CN111506423A (en) * | 2020-04-09 | 2020-08-07 | 网易（杭州）网络有限公司 | Resource scheduling method and device, electronic equipment and storage medium |
| CN112667403A (en) * | 2020-12-31 | 2021-04-16 | 北京百度网讯科技有限公司 | Server scheduling method and device and electronic equipment |
| CN114828579A (en) * | 2022-05-07 | 2022-07-29 | 福建万芯科技有限公司 | Energy-saving control method of container data center and related equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110077795A1 (en) * | 2009-02-13 | 2011-03-31 | American Power Conversion Corporation | Data center control |
| CN102620378A (en) * | 2011-01-27 | 2012-08-01 | 国际商业机器公司 | Method and system for data center energy saving controlling |
| CN102855157A (en) * | 2012-07-19 | 2013-01-02 | 浪潮电子信息产业股份有限公司 | Method for comprehensively scheduling load of servers |
| CN103092316A (en) * | 2013-01-22 | 2013-05-08 | 浪潮电子信息产业股份有限公司 | Server power consumption management system based on data mining |
| CN203010859U (en) * | 2013-01-15 | 2013-06-19 | 中国工商银行股份有限公司 | Refrigeration and air conditioning system for computer room |
-
2013
- 2013-09-05 CN CN201310399815.7A patent/CN104423531A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110077795A1 (en) * | 2009-02-13 | 2011-03-31 | American Power Conversion Corporation | Data center control |
| CN102620378A (en) * | 2011-01-27 | 2012-08-01 | 国际商业机器公司 | Method and system for data center energy saving controlling |
| CN102855157A (en) * | 2012-07-19 | 2013-01-02 | 浪潮电子信息产业股份有限公司 | Method for comprehensively scheduling load of servers |
| CN203010859U (en) * | 2013-01-15 | 2013-06-19 | 中国工商银行股份有限公司 | Refrigeration and air conditioning system for computer room |
| CN103092316A (en) * | 2013-01-22 | 2013-05-08 | 浪潮电子信息产业股份有限公司 | Server power consumption management system based on data mining |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016197849A1 (en) * | 2015-06-09 | 2016-12-15 | 邻元科技（北京）有限公司 | Control method, cooling apparatus system, cooling apparatus controller, cooling tower system, cooling tower controller, water pump system, and water pump controller |
| CN108139942A (en) * | 2015-09-30 | 2018-06-08 | 华为技术有限公司 | The method of the end-to-end power efficiency modeling of data center |
| WO2018023496A1 (en) * | 2016-08-03 | 2018-02-08 | 深圳中兴力维技术有限公司 | Device racking management method and apparatus |
| CN106096874A (en) * | 2016-08-18 | 2016-11-09 | 中国联合网络通信集团有限公司 | The energy efficiency managing method of data center, device and system |
| CN109375994A (en) * | 2018-09-10 | 2019-02-22 | 西安交通大学 | Data center's task temperature prediction and dispatching method based on RBF neural |
| CN109189190B (en) * | 2018-10-16 | 2020-07-14 | 西安交通大学 | Data center heat management method based on temperature prediction |
| CN109189190A (en) * | 2018-10-16 | 2019-01-11 | 西安交通大学 | A kind of data center's thermal management method based on temperature prediction |
| CN109800066A (en) * | 2018-12-13 | 2019-05-24 | 中国科学院信息工程研究所 | A kind of data center's energy-saving scheduling method and system |
| CN109890176A (en) * | 2019-03-01 | 2019-06-14 | 北京慧辰资道资讯股份有限公司 | A kind of method and device based on artificial intelligence optimization's energy consumption of machine room efficiency |
| CN110188932A (en) * | 2019-05-20 | 2019-08-30 | 国核电力规划设计研究院有限公司 | Consumption of data center prediction technique based on evaluation optimization |
| CN111258398A (en) * | 2020-01-10 | 2020-06-09 | 安擎(天津)计算机有限公司 | AI server fan speed adjusting method, system, terminal device and storage medium |
| CN111506423A (en) * | 2020-04-09 | 2020-08-07 | 网易（杭州）网络有限公司 | Resource scheduling method and device, electronic equipment and storage medium |
| CN111506423B (en) * | 2020-04-09 | 2023-06-30 | 网易（杭州）网络有限公司 | Resource scheduling method and device, electronic equipment and storage medium |
| CN112667403A (en) * | 2020-12-31 | 2021-04-16 | 北京百度网讯科技有限公司 | Server scheduling method and device and electronic equipment |
| CN114828579A (en) * | 2022-05-07 | 2022-07-29 | 福建万芯科技有限公司 | Energy-saving control method of container data center and related equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104423531A (en) | Data center energy consumption scheduling method and data center energy consumption scheduling device | |
| US11076507B2 (en) | Methods and systems for managing facility power and cooling | |
| CN109800066B (en) | Energy-saving scheduling method and system for data center | |
| WO2021104096A1 (en) | Method and apparatus for task scheduling in container cloud environment, and server and storage apparatus | |
| EP1877950B1 (en) | Methods and systems for managing facility power and cooling | |
| US8131515B2 (en) | Data center synthesis | |
| US8639482B2 (en) | Methods and systems for managing facility power and cooling | |
| US7881910B2 (en) | Methods and systems for managing facility power and cooling | |
| CN109375994B (en) | Data center task temperature prediction and scheduling method based on RBF neural network | |
| EP3286501A1 (en) | Predictive building control system and method for optimizing energy use and thermal comfort for a building or network of buildings | |
| KR101762061B1 (en) | System and method for predicting energy consumption based on cloud | |
| CN111722540A (en) | Energy Internet digital twin simulation system and method | |
| CN102016754A (en) | Arrangement for operating a data center using building automation system interface | |
| Ajorlou et al. | Artificial bee colony algorithm for CONWIP production control system in a multi-product multi-machine manufacturing environment | |
| CN113348330B (en) | Management method and control method of air conditioning system, storage medium and control platform | |
| CN110245771B (en) | Demand prediction method, demand control method and system | |
| CN104317654A (en) | Data center task scheduling method based on dynamic temperature prediction model | |
| US20210149353A1 (en) | Building control system with automatic control problem formulation using building information model | |
| US9964933B2 (en) | System for distributing electrical energy over a cluster of electrical devices, method for distributing electrical energy over a cluster of electrical devices of such a system and controller for use in such a system | |
| CN110175778A (en) | Unit efficiency planning system | |
| US20130275097A1 (en) | Air flow estimation | |
| US9459129B2 (en) | Determining sensor placement in distributed networks | |
| Marantos et al. | Rapid prototyping of low-complexity orchestrator targeting cyberphysical systems: The smart-thermostat usecase | |
| Vorozhtsov et al. | Resource control system stability of mobile data centers | |
| CN109460301A (en) | A kind of the flexible resource configuration method and system of flow data load |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150318 |
|
| RJ01 | Rejection of invention patent application after publication |