CN115907202B - Data center PUE (physical distribution element) calculation analysis method and system under double-carbon background - Google Patents

Data center PUE (physical distribution element) calculation analysis method and system under double-carbon background Download PDF

Info

Publication number
CN115907202B
CN115907202B CN202211601036.6A CN202211601036A CN115907202B CN 115907202 B CN115907202 B CN 115907202B CN 202211601036 A CN202211601036 A CN 202211601036A CN 115907202 B CN115907202 B CN 115907202B
Authority
CN
China
Prior art keywords
data center
energy consumption
function
pue
efficiency
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.)
Active
Application number
CN202211601036.6A
Other languages
Chinese (zh)
Other versions
CN115907202A (en
Inventor
原大伟
关朝阳
宋四海
冯臻
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China International Telecommunication Construction Group Design Institute Co ltd
Original Assignee
China International Telecommunication Construction Group Design Institute Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China International Telecommunication Construction Group Design Institute Co ltd filed Critical China International Telecommunication Construction Group Design Institute Co ltd
Priority to CN202211601036.6A priority Critical patent/CN115907202B/en
Publication of CN115907202A publication Critical patent/CN115907202A/en
Application granted granted Critical
Publication of CN115907202B publication Critical patent/CN115907202B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a data center PUE (physical unclonable element) calculation and analysis method and system under a double-carbon background, and relates to the technical field of data center management. According to the invention, the corresponding ageing correction quantity and the corresponding efficiency correction quantity are respectively constructed through the data center ageing correction function construction and the data center efficiency correction function construction; the aging correction can correct nonsensical decline of the PUE value caused by aging of IT equipment, and the effectiveness correction can correct the problem of abnormally increased PUE value caused by calculation optimization; in addition, the embodiment also analyzes the PUE related historical data to obtain the instruction for effective IT equipment aging maintenance and area optimization.

Description

Data center PUE (physical distribution element) calculation analysis method and system under double-carbon background
Technical Field
The invention relates to the technical field of data center management, in particular to a data center PUE (physical unclonable element) calculation and analysis method and system under a double-carbon background.
Background
PUE, power Usage Effectiveness, i.e., electrical energy utilization efficiency, is an indicator of evaluating the energy efficiency of a data center, and is the ratio of all energy consumed by the data center to the energy consumed by IT loads. PUE = data center total power consumption/IT equipment power consumption, wherein the data center total power consumption includes the power consumption of the IT equipment power consumption and the power consumption of the system such as refrigeration, distribution, etc., and ITs value is greater than 1, and the closer 1 indicates that the less the non-IT equipment power consumption is, the better the energy efficiency level is.
The existing PUE calculation and analysis method can theoretically promote the reduction of the energy consumption of the infrastructure of the data center, but as the index can not provide information about the energy efficiency of the IT equipment, the evaluation result and the actual production efficiency of the data center have great deviation, which is also an important reason for the problem caused by the PUE, and the point to be improved is mainly as follows:
1. the existing PUE theory only considers the duty ratio of the energy consumption of IT equipment in the total energy consumption of a data center; after the data center for carrying out infrastructure upgrade operates for a period of time, the energy consumption of the IT equipment can be increased due to the aging of the IT equipment; thereby increasing the IT device power consumption duty cycle so that the PUE value will decrease. However, the reduction of the PUE value is not from optimization of energy consumption of the data center, and cannot measure the power utilization efficiency of the data center, and if the aging energy consumption is continuously increased, the energy consumption is greatly wasted, and the potential risk of operation of the data center is increased.
2. The existing PUE theory does not consider the efficiency relation between IT equipment energy consumption and the total energy consumption of a data center; when IT devices adopt computing optimization technologies such as virtualization, the number of IT devices and the energy consumption thereof can be greatly reduced due to the increase of the working efficiency of a single device, which is beneficial to a data center. However, since the total energy consumption of the data center is not simply linearly related to the energy consumption of the IT equipment, the total energy consumption is not reduced to the same extent, which causes the PUE value to be increased instead.
3. The calculation definition formula of the existing PUE theory does not introduce a time concept, and cannot provide a substantial energy-saving direction for a data center.
Therefore, it is necessary to provide a method and a system for calculating and analyzing PUE of a data center in a dual-carbon background to solve the above technical problems.
Disclosure of Invention
In order to solve one of the technical problems, according to the data center PUE calculation and analysis method under the dual-carbon background, an optimized PUE model is constructed to calculate and adjust the PUE value, so that the optimized PUE value which is closer to the real situation of the data center is obtained; tracking the optimized PUE value and recording the PUE related historical data; performing optimization analysis according to the PUE related historical data to obtain an optimization improvement strategy; the method comprises the following specific steps: an optimized PUE calculation model construction step, an optimized PUE value tracking and recording step and an optimized PUE improvement analysis step.
As a still further solution, the optimizing PUE calculation model building step is for, in particular, the substeps comprising: constructing a data center aging correction function, constructing a data center efficiency correction function, integrating a PUE model construction and calculating a PUE model numerical value;
the data center aging correction function is constructed by: the aging correction function is used for evaluating the aging degree of the data center and constructing an aging correction function corresponding to the IT equipment; when specific calculation is carried out, the corresponding data center aging correction quantity is obtained by calculating the aging correction function of the IT equipment;
the data center efficiency correction function is constructed by: the method is used for evaluating the performance improvement of the optimization means on the data center and constructing a performance correction function corresponding to the IT equipment; when specific calculation is carried out, the corresponding IT equipment efficiency correction quantity is obtained by calculating the efficiency correction function of the IT equipment;
the PUE model correction function is integrated: integrating the aging correction function and the efficiency correction function into an original PUE calculation model to obtain an optimized PUE calculation model PUE +
The optimized PUE calculation model is calculated by numerical value: and (5) inputting specific values of each parameter of the optimized PUE calculation model to obtain an optimized PUE value.
As a further solution, the data center aging correction function construction specific steps include: setting energy consumption function acquisition conditions, acquiring an initial energy consumption function, acquiring a current energy consumption function and calculating an aging correction function;
the energy consumption function acquisition condition is set: setting unit sampling time t, setting sampling point number N and setting maximum working intensity W of data center max Setting a working strength gradient value delta W; wherein Δw=w max N; the working strength is defined as the ratio of the calculated amount of the data center in unit time, wherein the unit time is the unit sampling time t;
the initial energy consumption function is collected: under the energy consumption function acquisition condition, carrying out initial energy consumption test on the data center in the initial stage to obtain a plurality of initial energy consumption data sampling points; connecting all the initial energy consumption data sampling points to obtain an initial energy consumption curve; fitting the initial energy consumption curve to obtain an initial energy consumption function IDF;
the current energy consumption function is acquired: under the energy consumption function acquisition condition, carrying out current energy consumption test on the data center in the current stage to obtain a plurality of current energy consumption data sampling points; connecting all the current energy consumption data sampling points to obtain a current energy consumption curve; fitting the current energy consumption curve to obtain a current energy consumption function CDF;
the aging correction function is calculated: collecting an initial energy consumption function IDF and a current energy consumption function CDF, and calculating an aging correction function ADF; the aging correction function ADF is a difference function between the current energy consumption function CDF and the initial energy consumption function IDF.
As a still further solution, the initial energy consumption function is collected: setting the working strength of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center working strength values W according to the working strength gradient value delta W i Obtaining a working strength point set W; collecting to obtain W i Corresponding data center initial working energy consumption value E0 i Obtaining an initial working energy consumption point set E0; wherein W is i =ΔW*i,i∈[1,2,3....N];W=[W 1 ,W 2 ,W 3 ....W N ];E0=[E0 1 ,E0 2 ,E0 3 ....E0 N ]The method comprises the steps of carrying out a first treatment on the surface of the Converting E0 and W from the point set into a corresponding curve E0 (W), and obtaining an initial energy consumption function IDF through fitting: f (F) IDF (W)=E0(W);
The current energy consumption function is acquired: setting the working strength of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center working strength values W according to the working strength gradient value delta W i Obtaining a working strength point set W; collecting to obtain W i Corresponding data center current working energy consumption value E1 i Obtaining a current working energy consumption point set E1; wherein W is i =ΔW*i,i∈[1,2,3....N];W=[W 1 ,W 2 ,W 3 ....W N ];E1=[E1 1 ,E1 2 ,E1 3 ....E1 N ]The method comprises the steps of carrying out a first treatment on the surface of the Converting E1 and W from the point set into a corresponding curve E1 (W), and obtaining a current energy consumption function IDF through fitting: f (F) CDF (W)=E1(W);
The aging correction function is calculated: f (F) ADF (W)=F CDF (W)-F IDF (W)=E1(W)-E0(W)。
As a further solution, the data center efficiency correction function construction specific steps include: setting efficiency function acquisition conditions: the method comprises the steps of original efficiency function collection, optimized efficiency function collection and efficiency correction function calculation;
the efficiency function acquisition condition is set: setting a reference computing force u, setting the number M of working energy consumption sampling points and setting a working energy consumption boundary value E max Setting a calculated amount gradient value delta d; wherein Δd=e max M; the working energy consumption is the electric energy consumed by completing the appointed calculation amount under the current calculation force, wherein the current calculation force is the reference calculation force u;
the original efficacy function is collected: under the condition of efficiency function acquisition, closing calculation optimization; performing original efficiency test on the data center in the same test stage to obtain a plurality of original efficiency sampling points; connecting the original efficiency sampling points to obtain an original efficiency curve; fitting the original efficiency curve to obtain an original efficiency function OEF;
the optimized efficiency function is collected: under the condition of collecting the efficiency function, starting calculation optimization; performing optimization efficiency test on the data center in the same test stage to obtain a plurality of optimization efficiency sampling points; connecting the optimized efficiency sampling points to obtain an optimized efficiency curve; fitting the optimized efficiency curve to obtain an optimized efficiency function MEF;
the efficiency correction function is calculated: collecting original efficiency function OEF and optimized efficiency function MEF, and calculating efficiency correction function CEF; the efficiency correction function CEF is a difference function between the optimized efficiency function MEF and the original efficiency function OEF.
As a still further solution, the raw efficacy function is collected: setting the calculated amount of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center calculation amounts d according to the calculation amount gradient value delta d j Obtaining a workload point set d; collecting reference calculation force u to finish the alignmentCalculation amount d of response data center j Data center original working energy consumption value E of (2) j Obtaining an original working energy consumption point set E; wherein d j =Δd*j,i∈[1,2,3....M];d=[d 1 ,d 2 ,d 3 ....d m ];E=[E 1 ,E 2 ,E 3 ....E m ]The method comprises the steps of carrying out a first treatment on the surface of the E and d are converted into a corresponding curve E (d) from the point set, and an original efficacy function OEF is obtained through fitting: f (F) OEF (d)=E(d);
The optimized efficiency function is collected: setting the calculated amount of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center calculation amounts d according to the calculation amount gradient value delta d j Obtaining a workload point set d; acquiring reference computing power u to complete corresponding data center computing quantity d j Is used for optimizing the working energy consumption value of a data centerObtaining an optimized working energy consumption point set E'; wherein d j =Δd*j,i∈[1,2,3....M];d=[d 1 ,d 2 ,d 3 ....d m ];Converting E 'and d from the point set into a corresponding curve E' (d), and obtaining an optimized efficiency function MEF through fitting: f (F) MEF (d)=E`(d);
The efficiency correction function is calculated: f (F) CEF (d)=F MEF (d)-F OEF (d)=E`(d)-E(d)。
As a further solution, the PUE model correction function is integrated to obtain an optimized PUE calculation model PUE +
Wherein D, U and T respectively represent total calculation amount, total calculation force and total calculation time for PUE calculation statistics, D is the same as D units, U is the same as U units, and T is the same as T units; e (E) Tot Total energy consumption of the data center; e (E) IT For the total energy consumption of the IT equipment,a data center aging correction amount; />For the average working strength of the PUE calculation, < >>Is an IT device performance modifier.
As a still further solution, the PUE value trace recording step is for saving PUE-related history data including: an aging correction function history set and a region-optimized PUE value set.
As a still further solution, the optimizing PUE improvement analysis step includes: and (3) performing IT equipment aging maintenance analysis and data center area optimization analysis, wherein:
the IT equipment aging maintenance analysis comprises the following steps: reading an aging correction function history set, and setting an aging maintenance judgment threshold; under the unified working strength W, the data center aging correction quantity collected by each history node is calculated to obtain an aging correction quantity history Set (F) ADF )=[F ADF (W) 1 ,F ADF (W) 2 ,F ADF (W) 3 ......F ADF (W) k ]Wherein k is the node number; calculate the aging correction delta Set (DeltaF) ADF )=[F ADF (W) 1 -F ADF (W) 2 ,F ADF (W) 2 -F ADF (W) 3 ,......F ADF (W) k-1 -F ADF (W) k ]The method comprises the steps of carrying out a first treatment on the surface of the Determining an aging correction delta Set (ΔF) ADF ) Whether there is an element exceeding the aging maintenance determination threshold; if the information exists, the corresponding historical node IT equipment ages too fast, and reinforcing checking and maintenance are needed; if the equipment is not in the ageing speed, the equipment is properly maintained, and the normal ageing speed of the equipment is met;
the data center region optimization analysis: reading areaOptimizing a set of PUE values Wherein k is the region number; optimizing a Set of PUE values for a region (PUE + ) Sequencing from large to small to obtain a region optimization sequence; and carrying out partition optimization on the data center according to the region optimization sequence.
A data center PUE calculation and analysis system in a two-carbon background is deployed in a data center, and performs a data center PUE calculation and analysis method in a two-carbon background as described in any one of the above.
Compared with the related art, the data center PUE calculation and analysis method and system under the dual-carbon background have the following beneficial effects:
1. according to the invention, the corresponding ageing correction quantity and the corresponding efficiency correction quantity are respectively constructed through the data center ageing correction function construction and the data center efficiency correction function construction; the aging correction can correct nonsensical decline of the PUE value caused by aging of IT equipment, and the effectiveness correction can correct the problem of abnormally increased PUE value caused by calculation optimization;
2. according to the invention, the aging degree can be judged by the aging correction quantity of the data center collected by each history node and solving the corresponding aging correction increment set; if any aging correction increment is larger than the aging maintenance judgment threshold value, the fact that the aging speed of IT equipment is too high between corresponding historical nodes is indicated, so that larger aging correction quantity is needed to correct the IT equipment, and people can guide the maintenance of our daily equipment through the point;
3. the invention divides the data center into a plurality of areas and numbers the areas; each independent area carries out optimization PUE value calculation and sequencing from big to small to obtain an area optimization sequence; when the PUE value of the data center is optimized, according to the region optimization sequence, the region with the greatest influence on the PUE value of the data center can be processed most preferentially, and the PUE value of the data center can be optimized.
Drawings
FIG. 1 is a schematic diagram of a general flow chart of a method and a system for computing and analyzing PUE of a data center in a dual-carbon background according to an embodiment of the present invention;
FIG. 2 is a graph showing the comparison of energy consumption functions of a method and a system for computing and analyzing PUE of a data center in a dual-carbon background according to an embodiment of the present invention;
fig. 3 is a comparison graph of performance functions of a data center PUE calculation and analysis method and system in a dual-carbon background according to an embodiment of the present invention.
Detailed Description
The invention will be further described with reference to the drawings and embodiments.
As shown in fig. 1, in the data center PUE calculation and analysis method under a dual-carbon background provided by the embodiment, an optimized PUE model is constructed to calculate and adjust a PUE value, so as to obtain an optimized PUE value closer to the real situation of the data center; tracking the optimized PUE value and recording the PUE related historical data; performing optimization analysis according to the PUE related historical data to obtain an optimization improvement strategy; the method comprises the following specific steps: an optimized PUE calculation model construction step, an optimized PUE value tracking and recording step and an optimized PUE improvement analysis step.
As a still further solution, the optimizing PUE calculation model building step is for, in particular, the substeps comprising: constructing a data center aging correction function, constructing a data center efficiency correction function, integrating a PUE model construction and calculating a PUE model numerical value;
the data center aging correction function is constructed by: the aging correction function is used for evaluating the aging degree of the data center and constructing an aging correction function corresponding to the IT equipment; when specific calculation is carried out, the corresponding data center aging correction quantity is obtained by calculating the aging correction function of the IT equipment;
the data center efficiency correction function is constructed by: the method is used for evaluating the performance improvement of the optimization means on the data center and constructing a performance correction function corresponding to the IT equipment; when specific calculation is carried out, the corresponding IT equipment efficiency correction quantity is obtained by calculating the efficiency correction function of the IT equipment;
the PUE model correction function is integrated: integrating the aging correction function and the efficiency correction function into an original PUE calculation model to obtain an optimized PUE calculation model PUE +
The optimized PUE calculation model is calculated by numerical value: and (5) inputting specific values of each parameter of the optimized PUE calculation model to obtain an optimized PUE value.
It should be noted that: the existing PUE calculation method only considers the ratio of the IT equipment energy consumption to the total energy consumption of the data center, does not consider the efficiency relation between the IT equipment energy consumption and the total energy consumption of the data center, and does not introduce a time concept into the calculation definition type; therefore, optimization improvement is needed; the embodiment provides a data center PUE calculation and analysis method under a double-carbon background to solve the problem; respectively constructing corresponding aging correction quantity and efficiency correction quantity through data center aging correction function construction and data center efficiency correction function construction; the aging correction can correct nonsensical decline of the PUE value caused by aging of IT equipment, and the effectiveness correction can correct the problem of abnormally increased PUE value caused by calculation optimization; in addition, the embodiment also analyzes the PUE related historical data to obtain the instruction for effective IT equipment aging maintenance and area optimization.
As a further solution, the data center aging correction function construction specific steps include: setting energy consumption function acquisition conditions, acquiring an initial energy consumption function, acquiring a current energy consumption function and calculating an aging correction function;
the energy consumption function acquisition condition is set: setting unit sampling time t, setting sampling point number N and setting maximum working intensity W of data center max Setting a working strength gradient value delta W; wherein Δw=w max N; the working strength is defined as the ratio of the calculated amount of the data center in unit time, wherein the unit time is the unit sampling time t;
the initial energy consumption function is collected: under the energy consumption function acquisition condition, carrying out initial energy consumption test on the data center in the initial stage to obtain a plurality of initial energy consumption data sampling points; connecting all the initial energy consumption data sampling points to obtain an initial energy consumption curve; fitting the initial energy consumption curve to obtain an initial energy consumption function IDF;
the current energy consumption function is acquired: under the energy consumption function acquisition condition, carrying out current energy consumption test on the data center in the current stage to obtain a plurality of current energy consumption data sampling points; connecting all the current energy consumption data sampling points to obtain a current energy consumption curve; fitting the current energy consumption curve to obtain a current energy consumption function CDF;
the aging correction function is calculated: collecting an initial energy consumption function IDF and a current energy consumption function CDF, and calculating an aging correction function ADF; the aging correction function ADF is a difference function between the current energy consumption function CDF and the initial energy consumption function IDF.
It should be noted that: because the energy consumption of the IT equipment is increased due to the aging of the IT equipment, and the relationship of the increase is not simply linear, the initial energy consumption function needs to be acquired and then compared with the current energy consumption function, so that an aging correction function is obtained; the energy consumption increasing value brought by aging under different working strengths can be obtained through the aging correction function, and the influence of the energy consumption increasing value on the calculation of the PUE value can be corrected by taking the energy consumption increasing value as the aging correction quantity.
As shown in fig. 2, we can clearly see that, in standby, the initial energy consumption function IDF and the current energy consumption function CDF have standby initial values, but the energy consumption is not very different; however, as the working strength increases, the load on the equipment increases, and the energy consumption effect due to aging increases dramatically, so that it is necessary to remove the effect when calculating the PUE value.
As a still further solution, the initial energy consumption function is collected: setting the working strength of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center working strength values W according to the working strength gradient value delta W i Obtaining a working strength point set W; collecting to obtain W i Corresponding data center initial working energy consumption value E0 i Obtaining an initial working energy consumption point set E0; wherein W is i =ΔW*i,i∈[1,2,3....N];W=[W 1 ,W 2 ,W 3 ....W N ];E0=[E0 1 ,E0 2 ,E0 3 ....E0 N ]The method comprises the steps of carrying out a first treatment on the surface of the Converting E0 and W from the point set into a corresponding curve E0 (W), and obtaining an initial energy consumption function IDF through fitting: f (F) IDF (W)=E0(W);
The current energy consumption function is acquired: setting the working strength of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center working strength values W according to the working strength gradient value delta W i Obtaining a working strength point set W; collecting to obtain W i Corresponding data center current working energy consumption value E1 i Obtaining a current working energy consumption point set E1; wherein W is i =ΔW*i,i∈[1,2,3....N];W=[W 1 ,W 2 ,W 3 ....W N ];E1=[E1 1 ,E1 2 ,E1 3 ....E1 N ]The method comprises the steps of carrying out a first treatment on the surface of the Converting E1 and W from the point set into a corresponding curve E1 (W), and obtaining a current energy consumption function IDF through fitting: f (F) CDF (W)=E1(W);
The aging correction function is calculated: f (F) ADF (W)=F CDF (W)-F IDF (W)=E1(W)-E0(W)。
It should be noted that: since the function operation is required, the external conditions except the control variable and the observation variable are the same, and the energy consumption function acquisition conditions are unified.
As a further solution, the data center efficiency correction function construction specific steps include: setting efficiency function acquisition conditions: the method comprises the steps of original efficiency function collection, optimized efficiency function collection and efficiency correction function calculation;
the efficiency function acquisition condition is set: setting a reference computing force u, setting the number M of working energy consumption sampling points and setting a working energy consumption boundary value E max Setting a calculated amount gradient value delta d; wherein Δd=e max M; the working energy consumption is the electric energy consumed by completing the appointed calculation amount under the current calculation force, wherein the current calculation force is the reference calculation force u;
the original efficacy function is collected: under the condition of efficiency function acquisition, closing calculation optimization; performing original efficiency test on the data center in the same test stage to obtain a plurality of original efficiency sampling points; connecting the original efficiency sampling points to obtain an original efficiency curve; fitting the original efficiency curve to obtain an original efficiency function OEF;
the optimized efficiency function is collected: under the condition of collecting the efficiency function, starting calculation optimization; performing optimization efficiency test on the data center in the same test stage to obtain a plurality of optimization efficiency sampling points; connecting the optimized efficiency sampling points to obtain an optimized efficiency curve; fitting the optimized efficiency curve to obtain an optimized efficiency function MEF;
the efficiency correction function is calculated: collecting original efficiency function OEF and optimized efficiency function MEF, and calculating efficiency correction function CEF; the efficiency correction function CEF is a difference function between the optimized efficiency function MEF and the original efficiency function OEF.
It should be noted that: the construction method of the efficiency correction function is similar to that of the aging correction function, except that the variables are different, the acquisition conditions of the efficiency function are different, and the PUE value caused by correcting the efficiency deviation can be abnormally increased through the different conditions.
As shown in fig. 3, the original performance function and the optimized performance function have little difference in energy consumption when no calculation task is performed, but the two energy consumption is gradually separated with the increase of the calculation amount; at the boundary value E of energy consumption max When the calculation task limit of the calculation unit is reached, it is easy to see that the calculation amount after calculation optimization is much larger than that of non-optimization calculation, and the influence of the calculation amount is that the PUE value is increased when the PUE value is calculated, so that the influence needs to be eliminated.
As a still further solution, the raw efficacy function is collected: setting the calculated amount of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center calculation amounts d according to the calculation amount gradient value delta d j Obtaining a workload point set d; acquiring reference computing power u to complete corresponding data center computing quantity d j Data center original working energy consumption value E of (2) j Obtaining an original working energy consumption point set E; wherein d j =Δd*j,i∈[1,2,3....M];d=[d 1 ,d 2 ,d 3 ....d m ];E=[E 1 ,E 2 ,E 3 ....E m ]The method comprises the steps of carrying out a first treatment on the surface of the E and d are converted into a corresponding curve E (d) from the point set, and an original efficacy function OEF is obtained through fitting: f (F) OEF (d)=E(d);
The optimized efficiency function is collected: setting the calculated amount of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center calculation amounts d according to the calculation amount gradient value delta d j Obtaining a workload point set d; acquiring reference computing power u to complete corresponding data center computing quantity d j Is used for optimizing the working energy consumption value of a data centerObtaining an optimized working energy consumption point set E'; wherein d j =Δd*j,i∈[1,2,3....M];d=[d 1 ,d 2 ,d 3 ....d m ];Converting E 'and d from the point set into a corresponding curve E' (d), and obtaining an optimized efficiency function MEF through fitting: f (F) MEF (d)=E`(d);
The efficiency correction function is calculated: f (F) CEF (d)=F MEF (d)-F OEF (d)=E`(d)-E(d)。
As a further solution, the PUE model correction function is integrated to obtain an optimized PUE calculation model PUE +
Wherein D, U and T respectively represent total calculation amount, total calculation force and total calculation time for PUE calculation statistics, D is the same as D units, U is the same as U units, and T is the same as T units; e (E) Tot Total energy consumption of the data center; e (E) IT For the total energy consumption of the IT equipment,a data center aging correction amount; />For the average working strength of the PUE calculation, < >>Is an IT device performance modifier.
It should be noted that: the calculation is based on the same units, and if the units are different, normalization processing is also required, and the obtained distortion may be different, but the result is the same.
As a still further solution, the PUE value trace recording step is for saving PUE-related history data including: an aging correction function history set and a region-optimized PUE value set.
As a still further solution, the optimizing PUE improvement analysis step includes: and (3) performing IT equipment aging maintenance analysis and data center area optimization analysis, wherein:
the IT equipment aging maintenance analysis comprises the following steps: reading an aging correction function history set, and setting an aging maintenance judgment threshold; under the unified working strength W, the data center aging correction quantity collected by each history node is calculated to obtain an aging correction quantity history Set (F) ADF )=[F ADF (W) 1 ,F ADF (W) 2 ,F ADF (W) 3 ......F ADF (W) k ]Wherein k is the node number; calculate the aging correction delta Set (DeltaF) ADF )=[F ADF (W) 1 -F ADF (W) 2 ,F ADF (W) 2 -F ADF (W) 3 ,......F ADF (W) k-1 -F ADF (W) k ]The method comprises the steps of carrying out a first treatment on the surface of the Determining an aging correction delta Set (ΔF) ADF ) Whether there is an element exceeding the aging maintenance determination threshold; if the information exists, the corresponding historical node IT equipment ages too fast, and reinforcing checking and maintenance are needed; if not, then the methodProper maintenance is performed, and the normal aging speed of the equipment is met;
it should be noted that: according to the embodiment, the aging degree can be judged by the aging correction quantity of the data center collected by each history node and solving the corresponding aging correction increment set; if any aging correction increment is larger than the aging maintenance judging threshold value, the fact that the aging speed of IT equipment is too high between corresponding historical nodes is indicated, so that larger aging correction quantity is needed to correct the IT equipment, and people can guide the maintenance of our daily equipment through the point. Such as: and when the aging speed of IT equipment is found to be too high recently, the line inspection is carried out on each equipment of the data center, so that the increase of energy consumption and the elimination of potential risks are avoided.
The data center region optimization analysis: read zone optimized PUE value set Wherein k is the region number; optimizing a Set of PUE values for a region (PUE + ) Sequencing from large to small to obtain a region optimization sequence; and carrying out partition optimization on the data center according to the region optimization sequence.
It should be noted that: the data center is divided into a plurality of areas and numbered; each independent area carries out optimization PUE value calculation and sequencing from big to small to obtain an area optimization sequence; when the PUE value of the data center is optimized, according to the region optimization sequence, the region with the greatest influence on the PUE value of the data center can be processed most preferentially, and the PUE value of the data center can be optimized.
A data center PUE calculation and analysis system in a two-carbon background is deployed in a data center, and performs a data center PUE calculation and analysis method in a two-carbon background as described in any one of the above.
The foregoing is only illustrative of the present invention and is not to be construed as limiting the scope of the invention, and all equivalent structures or equivalent flow modifications which may be made by the teachings of the present invention and the accompanying drawings or which may be directly or indirectly employed in other related art are within the scope of the invention.

Claims (4)

1. A data center PUE calculation analysis method under a double-carbon background is characterized in that an optimization PUE model is constructed to calculate and adjust a PUE value, so that an optimization PUE value which is closer to the real situation of the data center is obtained; tracking the optimized PUE value and recording the PUE related historical data; performing optimization analysis according to the PUE related historical data to obtain an optimization improvement strategy; the method comprises the following specific steps: an optimized PUE calculation model construction step, an optimized PUE value tracking and recording step and an optimized PUE improvement analysis step;
the specific substeps of the optimizing PUE calculation model construction step include: constructing a data center aging correction function, constructing a data center efficiency correction function, integrating a PUE model construction and calculating a PUE model numerical value;
the data center aging correction function is constructed by: the aging correction function is used for evaluating the aging degree of the data center and constructing an aging correction function corresponding to the IT equipment; when specific calculation is carried out, the corresponding data center aging correction quantity is obtained by calculating the aging correction function of the IT equipment;
the data center efficiency correction function is constructed by: the method is used for evaluating the performance improvement of the optimization means on the data center and constructing a performance correction function corresponding to the IT equipment; when specific calculation is carried out, the corresponding IT equipment efficiency correction quantity is obtained by calculating the efficiency correction function of the IT equipment;
the PUE model correction function is integrated: integrating the aging correction function and the efficiency correction function into an original PUE calculation model to obtain an optimized PUE calculation model PUE +
The PUE model numerical calculation: introducing specific values of each parameter of the optimized PUE calculation model to obtain an optimized PUE value;
the data center aging correction function construction specific steps comprise: setting energy consumption function acquisition conditions, acquiring an initial energy consumption function, acquiring a current energy consumption function and calculating an aging correction function;
the energy consumption function acquisition condition is set:setting unit sampling time t, setting sampling point number N and setting maximum working intensity W of data center max Setting a working strength gradient value delta W; wherein Δw=w max N; the working strength is defined as the ratio of the calculated amount of the data center in unit time, wherein the unit time is the unit sampling time t;
the initial energy consumption function is collected: under the energy consumption function acquisition condition, carrying out initial energy consumption test on the data center in the initial stage to obtain a plurality of initial energy consumption data sampling points; connecting all the initial energy consumption data sampling points to obtain an initial energy consumption curve; fitting the initial energy consumption curve to obtain an initial energy consumption function IDF;
the current energy consumption function is acquired: under the energy consumption function acquisition condition, carrying out current energy consumption test on the data center in the current stage to obtain a plurality of current energy consumption data sampling points; connecting all the current energy consumption data sampling points to obtain a current energy consumption curve; fitting the current energy consumption curve to obtain a current energy consumption function CDF;
the aging correction function is calculated: collecting an initial energy consumption function IDF and a current energy consumption function CDF, and calculating an aging correction function ADF; the aging correction function ADF is a difference function between the current energy consumption function CDF and the initial energy consumption function IDF;
the data center efficiency correction function construction specific steps comprise: setting efficiency function acquisition conditions: the method comprises the steps of original efficiency function collection, optimized efficiency function collection and efficiency correction function calculation;
the efficiency function acquisition condition is set: setting a reference computing force u, setting the number M of working energy consumption sampling points and setting a working energy consumption boundary value E max Setting a calculated amount gradient value delta d; wherein Δd=e max M; the working energy consumption is the electric energy consumed by completing the appointed calculation amount under the current calculation force, wherein the current calculation force is the reference calculation force u;
the original efficacy function is collected: under the condition of efficiency function acquisition, closing calculation optimization; performing original efficiency test on the data center in the same test stage to obtain a plurality of original efficiency sampling points; connecting the original efficiency sampling points to obtain an original efficiency curve; fitting the original efficiency curve to obtain an original efficiency function OEF;
the optimized efficiency function is collected: under the condition of collecting the efficiency function, starting calculation optimization; performing optimization efficiency test on the data center in the same test stage to obtain a plurality of optimization efficiency sampling points; connecting the optimized efficiency sampling points to obtain an optimized efficiency curve; fitting the optimized efficiency curve to obtain an optimized efficiency function MEF;
the efficiency correction function is calculated: collecting original efficiency function OEF and optimized efficiency function MEF, and calculating efficiency correction function CEF; the efficiency correction function CEF is a difference function between the optimized efficiency function MEF and the original efficiency function OEF;
the step of tracking and recording the PUE value is used for storing the PUE related history data, and the PUE related history data comprises the following steps:
an aging correction function history set and a region optimization PUE value set;
the optimizing PUE improvement analysis step includes: and (3) performing IT equipment aging maintenance analysis and data center area optimization analysis, wherein:
the IT equipment aging maintenance analysis comprises the following steps: reading an aging correction function history set, and setting an aging maintenance judgment threshold; under the unified working strength W, the data center aging correction quantity collected by each history node is calculated to obtain an aging correction quantity history Set (F) ADF )=[F ADF (W) 1 ,F ADF (W) 2 ,F ADF (W) 3 ......F ADF (W) k ]Wherein k is the node number; calculate the aging correction delta Set (DeltaF) ADF )=[F ADF (W) 1 -F ADF (W) 2 ,F ADF (W) 2 -F ADF (W) 3 ,......F ADF (W) k-1 -F ADF (W) k ]The method comprises the steps of carrying out a first treatment on the surface of the Determining an aging correction delta Set (ΔF) ADF ) Whether there is an element exceeding the aging maintenance determination threshold; if the information exists, the corresponding historical node IT equipment ages too fast, and reinforcing checking and maintenance are needed; if the equipment is not in the ageing speed, the equipment is properly maintained, and the normal ageing speed of the equipment is met; which is a kind ofWherein the data center aging correction is calculated by an aging correction function as follows: f (F) ADF (W)=F CDF (W)-F IDF (W);F CDF (W) is the current energy consumption function, F IDF (W) is an initial energy consumption function;
the data center region optimization analysis: read zone optimized PUE value set Wherein k is the region number; optimizing a Set of PUE values for a region (PUE + ) Sequencing from large to small to obtain a region optimization sequence; and carrying out partition optimization on the data center according to the region optimization sequence.
2. A data center PUE computational analysis method in a two-carbon background according to claim 1, wherein,
the initial energy consumption function is collected: setting the working strength of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center working strength values W according to the working strength gradient value delta W i Obtaining a working strength point set W; collecting to obtain W i Corresponding data center initial working energy consumption value E0 i Obtaining an initial working energy consumption point set E0; wherein W is i =ΔW*i,i∈[1,2,3....N];W=[W 1 ,W 2 ,W 3 ....W N ];E0=[E0 1 ,E0 2 ,E0 3 ....E0 N ]The method comprises the steps of carrying out a first treatment on the surface of the Converting E0 and W from the point set into a corresponding curve E0 (W), and obtaining an initial energy consumption function IDF through fitting: f (F) IDF (W)=E0(W);
The current energy consumption function is acquired: setting the working strength of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center working strength values W according to the working strength gradient value delta W i Obtaining a working strength point set W; collecting to obtain W i Corresponding number ofAccording to the current working energy consumption value E1 of the center i Obtaining a current working energy consumption point set E1; wherein W is i =ΔW*i,i∈[1,2,3....N];W=[W 1 ,W 2 ,W 3 ....W N ];E1=[E1 1 ,E1 2 ,E1 3 ....E1 N ]The method comprises the steps of carrying out a first treatment on the surface of the Converting E1 and W from the point set into a corresponding curve E1 (W), and obtaining a current energy consumption function IDF through fitting: f (F) CDF (W)=E1(W)。
3. A data center PUE computational analysis method in a two-carbon background according to claim 2, wherein,
the original efficacy function is collected: setting the calculated amount of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center calculation amounts d according to the calculation amount gradient value delta d j Obtaining a workload point set d; acquiring reference computing power u to complete corresponding data center computing quantity d j Data center original working energy consumption value E of (2) j Obtaining an original working energy consumption point set E; where dj=Δd×j, j E [1,2,3 ] M];d=[d 1 ,d 2 ,d 3 ....d m ];E=[E 1 ,E 2 ,E 3 ....E m ]E and d are converted into a corresponding curve E (d) from the point set, and an original efficacy function OEF is obtained through fitting: f (F) OEF (d)=E(d);
The optimized efficiency function is collected: setting the calculated amount of the data center as a control variable, and setting the working energy consumption of the data center as an observation variable; setting different data center calculation amounts d according to the calculation amount gradient value delta d j Obtaining a workload point set d; acquiring reference computing power u to complete corresponding data center computing quantity d j Is used for optimizing the working energy consumption value E' of the data center j Obtaining an optimized working energy consumption point set E'; wherein d j =Δd*j,j∈[1,2,3…M];d=[d 1 ,d 2 ,d 3 ....d m ];E`=[E` 1 ,E` 2 E` 3 ....E` M ]Converting E 'and d from the point set into a corresponding curve E' (d), and obtaining an optimized efficiency function MEF through fitting: f (F) MEF (d)=E`(d);
The efficiency correction function is calculated: f (F) CEF (d)=F MEF (d)-F OEF (d)=E`(d)-E(d)。
4. The method for computing and analyzing a PUE of a data center in a two-carbon background as recited in claim 3, wherein said PUE model modification function is integrated to obtain an optimized PUE computing model PUE +
Wherein D, U and T respectively represent total calculation amount, total calculation force and total calculation time for PUE calculation statistics, D is the same as D units, U is the same as U units, and T is the same as T units; e (E) Tot Total energy consumption of the data center; e (E) IT For the total energy consumption of the IT equipment,a data center aging correction amount; />For the average working strength of the PUE calculation, < >>F CEF (D/U) is the IT device performance modifier.
CN202211601036.6A 2022-12-13 2022-12-13 Data center PUE (physical distribution element) calculation analysis method and system under double-carbon background Active CN115907202B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211601036.6A CN115907202B (en) 2022-12-13 2022-12-13 Data center PUE (physical distribution element) calculation analysis method and system under double-carbon background

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211601036.6A CN115907202B (en) 2022-12-13 2022-12-13 Data center PUE (physical distribution element) calculation analysis method and system under double-carbon background

Publications (2)

Publication Number Publication Date
CN115907202A CN115907202A (en) 2023-04-04
CN115907202B true CN115907202B (en) 2023-10-24

Family

ID=86481335

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211601036.6A Active CN115907202B (en) 2022-12-13 2022-12-13 Data center PUE (physical distribution element) calculation analysis method and system under double-carbon background

Country Status (1)

Country Link
CN (1) CN115907202B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102854418A (en) * 2012-08-27 2013-01-02 北京华胜天成科技股份有限公司 Energy usage effectiveness detector and detection system
CN109462223A (en) * 2018-04-09 2019-03-12 国网浙江省电力有限公司嘉兴供电公司 A kind of power quality and analysis method for reliability based on big data
CN111752710A (en) * 2020-06-23 2020-10-09 中国电力科学研究院有限公司 Data center PUE dynamic optimization method, system, equipment and readable storage medium
WO2020227983A1 (en) * 2019-05-15 2020-11-19 Alibaba Group Holding Limited Hybrid-learning neural network architecture
CN114810287A (en) * 2021-06-17 2022-07-29 长城汽车股份有限公司 Method and system for correcting LNT (Low-Density LNT) aging and vehicle

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8346398B2 (en) * 2008-08-08 2013-01-01 Siemens Industry, Inc. Data center thermal performance optimization using distributed cooling systems
US11461210B2 (en) * 2019-06-26 2022-10-04 Kyndryl, Inc. Real-time calculation of data center power usage effectiveness

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102854418A (en) * 2012-08-27 2013-01-02 北京华胜天成科技股份有限公司 Energy usage effectiveness detector and detection system
CN109462223A (en) * 2018-04-09 2019-03-12 国网浙江省电力有限公司嘉兴供电公司 A kind of power quality and analysis method for reliability based on big data
WO2020227983A1 (en) * 2019-05-15 2020-11-19 Alibaba Group Holding Limited Hybrid-learning neural network architecture
CN113518962A (en) * 2019-05-15 2021-10-19 阿里巴巴集团控股有限公司 Hybrid learning neural network architecture
CN111752710A (en) * 2020-06-23 2020-10-09 中国电力科学研究院有限公司 Data center PUE dynamic optimization method, system, equipment and readable storage medium
CN114810287A (en) * 2021-06-17 2022-07-29 长城汽车股份有限公司 Method and system for correcting LNT (Low-Density LNT) aging and vehicle

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
"数据中心PUE能效优化的机器学习方法";杨震 等;《***工程理论与实践》;第第42卷卷(第第3期期);全文 *
"数据中心能效指标与能耗模型研究";王少鹏 等;《信息通信技术与政策》(第第2期期);全文 *

Also Published As

Publication number Publication date
CN115907202A (en) 2023-04-04

Similar Documents

Publication Publication Date Title
CN114298863B (en) Data acquisition method and system of intelligent meter reading terminal
CN112901449B (en) Air compressor system energy consumption optimization method based on machine learning
CN112070353B (en) Method and system for accurately detecting energy efficiency of data center
CN116887569B (en) Data center energy consumption prediction and energy saving adjustment method, system and storage medium
CN112433907A (en) Method and device for processing host operation parameter data of uninterruptible power supply and electronic device
CN112365090A (en) Deep learning-based non-invasive electrical load identification method and device
CN112465239A (en) Desulfurization system operation optimization method based on improved PSO-FCM algorithm
CN117851908B (en) Improved on-line low-voltage transformer area electric energy meter misalignment monitoring method and device
CN110059938B (en) Power distribution network planning method based on association rule driving
CN111914000B (en) Server power capping method and system based on power consumption prediction model
CN117650628B (en) Energy efficiency management system based on self-adaptive preconditioning scene
CN115907202B (en) Data center PUE (physical distribution element) calculation analysis method and system under double-carbon background
CN116249186A (en) Data processing method and device of wireless network equipment, storage medium and electronic equipment
CN116470491A (en) Photovoltaic power probability prediction method and system based on copula function
CN110298767A (en) A kind of thermal power plant time series variable method for monitoring abnormality and system
CN108847666A (en) Analysis method, device and the realization device of power distribution network short circuit current level
Li et al. A fusion framework using integrated neural network model for non-intrusive load monitoring
CN116307285B (en) New energy development prediction device and method
CN117505074B (en) Electric precipitation control method and system based on energy-saving analysis
CN117691594B (en) Energy saving and consumption reduction judging method and system for transformer
CN110109974B (en) Die casting machine production data intelligent acquisition system based on power information
CN118051858B (en) Intelligent electricity saving method and system based on abnormal electricity consumption data
CN118300101A (en) Method, apparatus and storage medium for predicting electric load
CN118017502A (en) Digital twinning-based power distribution calculation power prediction method, system and medium
CN117591388A (en) Power equipment control method and system based on computing power architecture

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
GR01 Patent grant
GR01 Patent grant