CN110941396A - Copy placement method based on airflow organization and oriented to cloud data center - Google Patents

Copy placement method based on airflow organization and oriented to cloud data center Download PDF

Info

Publication number
CN110941396A
CN110941396A CN201911155244.6A CN201911155244A CN110941396A CN 110941396 A CN110941396 A CN 110941396A CN 201911155244 A CN201911155244 A CN 201911155244A CN 110941396 A CN110941396 A CN 110941396A
Authority
CN
China
Prior art keywords
disk
data center
energy consumption
disks
active
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
Application number
CN201911155244.6A
Other languages
Chinese (zh)
Inventor
邓玉辉
林瑞虹
范志峰
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.)
Jinan University
University of Jinan
Original Assignee
Jinan University
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 Jinan University filed Critical Jinan University
Priority to CN201911155244.6A priority Critical patent/CN110941396A/en
Publication of CN110941396A publication Critical patent/CN110941396A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
    • G06F3/0625—Power saving in storage systems
    • 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
    • G06F1/16—Constructional details or arrangements
    • G06F1/20—Cooling means
    • G06F1/206—Cooling means comprising thermal management
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
    • G06F3/0629—Configuration or reconfiguration of storage systems
    • G06F3/0631—Configuration or reconfiguration of storage systems by allocating resources to storage systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0668—Interfaces specially adapted for storage systems adopting a particular infrastructure
    • G06F3/0671—In-line storage system
    • G06F3/0683—Plurality of storage devices
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Power Sources (AREA)

Abstract

The invention discloses an energy-saving copy placement method based on airflow organization and oriented to a cloud data center, which constructs the relation between refrigeration energy consumption and disk energy consumption according to a thermal cycle model among data center nodes to obtain a total energy consumption model of the data center; calculating a thermal sensing disk sequence of the data center according to the obtained total energy consumption model; firstly storing an incoming request in a request queue, and then distributing a disk to which a copy belongs according to a thermal sensing disk sequence according to the request type; and calculating the distribution of the disks allocated with the requests in the current time period to obtain the refrigeration supply temperature of the data center, and adjusting the corresponding air conditioner cooling temperature. The invention provides an energy-saving copy placing method for a data center, and the method can control air-conditioning refrigeration equipment to work at higher temperature while reducing the energy consumption of a magnetic disk through online management of the magnetic disk data, thereby effectively reducing the total energy consumption of the data center.

Description

Copy placement method based on airflow organization and oriented to cloud data center
Technical Field
The invention relates to the technical field of copy placement in a data center, in particular to a cloud data center-oriented copy placement method based on airflow organization.
Background
With the development of the internet industry and the continuous generation of data, the data center stores user and enterprise data. The generation and storage of a large amount of data continuously enlarge the scale of the data center, the carbon emission and the operation cost are also continuously increased, and the energy consumption problem of the data center needs to be paid attention urgently.
Data is increased explosively, and much attention is paid to how to effectively reduce the energy consumption of the data center. For a data center, the existing service cannot be influenced while the energy consumption is reduced.
In the aspect of saving storage energy consumption, energy is saved mainly by adjusting the placement mode of data and the combination form of disks. For example, energy saving is achieved by an energy saving algorithm of the disk array, a data tilt placement method, and by analyzing the relevance of data to place a remote copy. However, these methods have poor effects in saving energy consumption and refrigeration energy consumption at the same time, which may cause a problem of local machine overheating or a problem of storage energy consumption due to over-dispersed storage, and are also complicated in computation amount, which may cause many limitations in online application.
Disclosure of Invention
The invention aims to provide an energy-saving copy placing method based on airflow organization for a cloud data center, aiming at the problems of huge operation cost and energy consumption and low energy efficiency ratio of the data center, so that the computational complexity of the traditional algorithm is improved, and the balance of storage energy consumption and refrigeration energy consumption is considered at the same time.
The purpose of the invention can be achieved by adopting the following technical scheme:
an energy-saving copy placing method based on airflow organization and oriented to a cloud data center is applied to data center copy placing and comprises the following steps:
constructing a relation between refrigeration energy consumption and disk energy consumption according to a thermal cycle model among data center nodes to obtain a data center total energy consumption model;
calculating a thermal sensing disk sequence of the data center according to the obtained total energy consumption model;
dividing the thermal sensing magnetic disk sequence according to the set copy number and the active copies;
firstly storing an incoming request in a request queue, and then distributing a disk to which a copy belongs according to a thermal sensing disk sequence according to the request type;
and calculating the distribution of the disks allocated with the requests in the current time period to obtain the refrigeration supply temperature of the data center, and adjusting the corresponding air conditioner cooling temperature.
Further, the process of constructing the relationship between the refrigeration energy consumption and the disk energy consumption according to the thermal cycle model among the data center nodes to obtain the total energy consumption model of the data center is as follows:
dividing the data center into a plurality of nodes according to the physical placement position of the disk, and modeling the air flow organization of the data center by adopting fluid dynamics CFD (computational fluid dynamics) to obtain energy consumption models of a disk storage system and an air-conditioning refrigeration system, thereby calculating the total energy consumption of the data center; and meanwhile, obtaining a model of the relationship between the total energy consumption of the data center and the number of the active disks in each node according to the model.
Further, in the process of dividing the disks into a plurality of nodes according to the physical placement positions of the disks, the disks which are in the same placement position area and share the power supply are classified into the same node according to the regionality of the physical placement positions and the power supply sharing state of the disks.
Further, in the process of modeling the airflow organization of the data center by adopting the fluid dynamics CFD, according to the divided region nodes, the fluid dynamics CFD is simulated, and the thermal airflow organization conduction coefficient between the nodes is calculated, so that a thermal cycle influence coefficient matrix is obtained.
Further, in the process of obtaining the model of the relationship between the total energy consumption of the data center and the number of the active disks in each node according to the model, the mathematical relationship between the storage cooling energy consumption and the storage energy consumption is calculated according to the obtained thermal cycle influence coefficient matrix, and then the total energy consumption of the data center is obtained, so that the model of the relationship between the total energy consumption of the data center and the number of the active disks in each node is obtained.
Further, in the process of calculating the thermal sensing disk sequence of the data center according to the obtained total energy consumption model, assuming that no active disk exists at a node in the data center at first, the number of the active disks is sequentially increased, and the node to which the position of the active disk belongs is started to be allocated, wherein the calculation method comprises the following steps:
t1, when the node to which the first active disk belongs is distributed, traversing all nodes, finding out the distributed node position with the lowest total data energy consumption in the corresponding energy consumption model, and recording the distributed node position;
t2, when the xth active disk is distributed, assuming that the previous x-1 disks are all fixed in the distributed positions in the previous step, searching the position with the lowest energy consumption of the data center corresponding to the xth disk, if the number of the disks already distributed by the node is equal to the number of the disks owned by the node, searching the distribution again, otherwise, recording the position of the distributed node, wherein the value of x is 2 to the maximum value of the number of the disks of the data center;
and T3, sequentially increasing the number of active disks and distributing according to the step T2 until the maximum number of disks are distributed to the data center, wherein all recorded disk distribution node positions form a heat sensing disk sequence.
Further, the process of dividing the thermal sensing disk sequence according to the set copy number and the active copy is as follows:
Figure BDA0002284631290000031
wherein, divider is the dividing length of the disk number sequence, copyactiveCopy number of active diskinactiveFor inactive disk copy number, disknumFloor is rounded down for the total number of disks.
Further, the incoming requests are firstly stored in a request queue, and then according to the request types, in the process of allocating the disks to which the copies belong according to the thermal sensing disk sequence, the request types are divided into read requests and write requests, the read requests require the disks in which the active copies are located to be online, for the write requests, if a data set already exists, the disks of all the copies are required to be online, otherwise, the copies are allocated according to the following method:
the allocation for the active copies is as follows:
s11, partitioning the active copies of the heat sensing disk sequence, sequentially inquiring whether the disk has a residual space for storing a data set from the head to the tail of the partitioned sequence, if the disk with the first space is inquired, successfully allocating, and recording the copy allocation by using a data allocation table;
s12, inquiring whether the disk has a residual space for storing a data set from the last found disk position of the partition to the tail of the partition, if the inquired disk with the space is the first disk, the allocation is successful, and recording the copy allocation by the data allocation table;
s13, repeating the previous step until all the required active copies are distributed;
the allocation for inactive copies is as follows:
s21, partitioning the inactive copies of the heat sensing disk sequence, sequentially inquiring whether the disk has a residual space for storing a data set from the head to the tail of the partitioned sequence, if the disk with the first space is inquired, successfully allocating, and recording the copy allocation by using a data allocation table;
s22, inquiring whether the disk has a residual space for storing a data set from the last found disk position of the partition to the tail of the partition, if the inquired disk with the space is the first disk, the allocation is successful, and recording the copy allocation by the data allocation table;
and S23, repeating the previous step until all the required active copies are distributed.
Further, the data center total energy consumption model is modeled according to the gas flow organization, and the following minimized energy consumption model is obtained:
Figure BDA0002284631290000051
Figure BDA0002284631290000052
Request＝(DataSet,RequestType)
Y＝f(DataTable,Request)
wherein the content of the first and second substances,
P＝prunY+pidleλ
Figure BDA0002284631290000053
Figure BDA0002284631290000054
the data center is assumed to have m nodes, Request represents a task Request, DataSet represents a data set number of the Request, RequestType represents a Request type and is divided into a READ type and a WRITE type, DataTable represents a storage position of a data set in the data center, Y represents the number of online disks of each node, and Y is used foriTo represent the number of active disks per node of the data center, and Y represents YiVector of composition, P is the computing device energy consumption vector, PiThe ith in the P vector represents the storage energy consumption of each node, COP is the coefficient of performance of the air-conditioning refrigeration equipment, and tsupCooling temperature, lambda, supplied to the air-conditioning unitiThe binary variable of 0 or 1 indicates whether the ith node is in an active state, 0 indicates no, 1 indicates yes, and lambda is lambdaiComposed vector, idle active nodes can consume pidleIs represented by prunThe temperature alarm value is t and represents the additionally increased energy consumption for starting each active disk in a single nodecriticalThe heat cycle matrix D represents the relationship between the coefficient of the heat mutual influence between the nodes and the power consumption of the nodes.
Compared with the prior art, the invention has the following advantages and effects:
(1) compared with a random placement strategy, the method has the advantages that under the condition of different read-write ratios, the performance is more energy-saving, and the energy consumption of the data center is reduced to a greater extent.
(2) Compared with a Hadoop-like copy placement method, the method has the advantages that under the condition of different read-write ratios, the performance is more energy-saving, and the energy consumption of a data center is reduced to a greater extent.
(3) Compared with the two strategies, the invention has the advantages that the times of disk awakening and sleep test are less in the disk awakening and sleep test evaluation, and the damage to the disk is reduced.
Drawings
FIG. 1 is a schematic diagram of the architecture of a data center replica allocator of the present invention;
FIG. 2 is a flow chart of a method for calculating a thermal-aware disk sequence according to the present invention;
FIG. 3 is a schematic diagram of active replica partitions, inactive replica partitions of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Examples
The embodiment discloses an energy-saving copy placing method based on airflow organization and oriented to a cloud data center, which comprises the following steps:
step P1, establishing an energy consumption model required by the calculation of the heat sensing disk sequence, firstly dividing the energy consumption model into a plurality of nodes according to the physical arrangement position of the disk, and modeling the air flow organization of the data center by adopting fluid dynamics CFD to obtain the energy consumption models of the disk storage system and the air-conditioning refrigeration system, thereby calculating the total energy consumption of the data center. And meanwhile, obtaining a model of the relationship between the total energy consumption of the data center and the number of the active disks in each node according to the model.
Dividing the disks into a plurality of nodes according to the physical placement positions of the disks, and classifying the disks which are in the same placement position area and share the power supply into the same node according to the regionality of the physical placement positions and the power supply sharing state of the disks.
And modeling the air flow organization of the data center by adopting fluid dynamics CFD (computational fluid dynamics), simulating the fluid dynamics CFD and calculating the conduction coefficient of the hot air flow organization between the nodes according to the divided region nodes to obtain a thermal cycle influence coefficient matrix.
And in a model for obtaining the relation between the total energy consumption of the data center and the number of the active disks in each node according to the model, calculating the mathematical relation between the storage cooling energy consumption and the storage energy consumption according to the obtained thermal cycle influence coefficient matrix, and further obtaining the total energy consumption of the data center to obtain a relation model between the total energy consumption of the data center and the number of the active disks in each node.
Assuming that no node in the data center has an active disk at first, sequentially increasing the number of the active disks, and starting to allocate the node to which the position of the active disk belongs, the calculation method comprises the following steps:
t1, when the node to which the first active disk belongs is distributed, traversing all nodes, finding out the distributed node position with the lowest total data energy consumption in the corresponding energy consumption model, and recording the distributed node position;
t2, when the xth active disk is distributed, assuming that the previous x-1 disks are all fixed in the positions distributed in the previous step, searching the position with the lowest energy consumption of the data center corresponding to the xth disk, if the number of the disks distributed by the node is equal to the number of the disks owned by the node, searching the distribution again, otherwise, recording the position of the distributed node, wherein the value of x is 2 to the maximum value of the number of the disks of the data center;
t3, sequentially increasing the number of active disks and allocating according to the step T2 until the maximum number of disks are allocated to the data center, and all recorded disk allocation node positions form a heat sensing disk sequence.
The data center total energy consumption model is modeled according to the gas flow organization, and the following minimized energy consumption model is obtained:
Figure BDA0002284631290000071
Figure BDA0002284631290000072
Request＝(DataSet,RequestType)
Y＝f(DataTable,Request)
wherein the content of the first and second substances,
P＝prunY+pidleλ
Figure BDA0002284631290000081
Figure BDA0002284631290000082
the data center is assumed to have m nodes, Request represents a task Request, DataSet represents a data set number of the Request, RequestType represents a Request type and is divided into a READ type and a WRITE type, DataTable represents a storage position of a data set in the data center, Y represents the number of online disks of each node, and Y is used foriTo represent the number of active disks per node of the data center, and Y represents YiVector of composition, P is the computing device energy consumption vector, PiThe ith in the P vector represents the storage energy consumption of each node, COP is the coefficient of performance of the air-conditioning refrigeration equipment, and tsupCooling temperature, lambda, supplied to the air-conditioning unitiThe binary variable of 0 or 1 indicates whether the ith node is in an active state, 0 indicates no, 1 indicates yes, and lambda is lambdaiComposed vector, idle active nodes can consume pidleIs represented by prunThe temperature alarm value is t and represents the additionally increased energy consumption for starting each active disk in a single nodecriticalThe heat cycle matrix D represents the relationship between the coefficient of the heat mutual influence between the nodes and the power consumption of the nodes.
As shown in fig. 1, the architecture of the data center replica placement machine of the present invention is shown, and the replica placement machine comprises:
1) request Queue (RQ, Request Queue)
This component is the entry point for the copy placer (RA) to process the read and write request stream. The method stores data requests to be processed in the next time period into a predefined sequence queue, and sends Read requests and Write requests to a Read Request Handler (RRH) and a Write Request Handler (WRH) respectively. The RQ also sends the access Data of the Data request to a Data Migration Handler (DMH) for statistics to help the DMH make decisions for hot spot Data migration.
2) Read Request Handler (RRH, Read Request Handler)
This component is primarily handling read requests. The method comprises the steps of acquiring a read request of a data set from an RQ, then inquiring a copy storage position corresponding to the data set from a data copy position record Table (RT, replay Table), and setting all active copies of the data set to be in an enabled state according to a strategy, namely setting a disk where the active copies of the data set are located to be in an online state. And sends information such as Online data set copies, disk numbers and the like to an Online disk decision maker (ODD, Online disks resolver). Then the data transmission between the read request and the disk is established to complete the read request.
3) Write Request Handler (WRH, Write Request Handler)
This component is primarily handling write requests. The method comprises the steps of obtaining a write request of a data set from RQ, then inquiring an idle disk space table from RT, obtaining an available space in an active copy partition of a disk according to a strategy, distributing an active copy of the data set to the space, obtaining the available space in an inactive copy partition of the disk according to the strategy, distributing an inactive copy of the data set to the space, and sending the placement position results of all the copies to the RT for recording. And all Disks needing disk space provision are set to be in an Online state so as to facilitate data writing, and information such as Online data set copies, disk numbers and the like is sent to an Online disk decision maker (ODD). Then the data transmission between the data storage device and the disk is established to complete the write request.
4) Data Migration processor (DMH, Data Migration Handler)
This component primarily performs data migration. It is contemplated that the amount of access to different data sets may vary greatly over time, such as where some new data is changing gradually to hot data, etc. The data migration is to set an access request statistics device of a data set, to perform statistics on the access request metadata transmitted from the RQ at intervals according to a strategy, to acquire the distribution of the current data from the RT, and to determine whether to migrate some data sets according to the statistics and the distribution of the current data.
5) Data copy position record Table (RT, Replica Table)
This component is primarily the storage location where the data set copy is recorded. The RT provides the RRH, WRH and DMH with the number and location of copies of the query data set, the free space of the query disk, and records the results of the placement and migration of the copies of the data set obtained from the WRH and DMH.
6) Online disk Decider (ODD, on Disks resolver)
This component determines which disks are online for a certain period of time. And the ODD acquires the copy of the data set and the disk position information which need to be online from the RRH and the DMH, and finally determines the online disk distribution in the next time period according to a strategy.
7) Supply Temperature Setter (STS)
This component collects real-time temperature information of the Computer Room cooling equipment (CRAC) and receives ideal temperature information that can be currently set calculated by the TAC. The optimal supply temperature is then determined and sent to the CRAC.
Step P2, calculating a data center thermal sensing disk sequence according to the established energy consumption model;
as shown in fig. 2, according to the energy consumption model, assuming that no active disk exists at a node in the data center at first, sequentially increasing the number of active disks, and starting to allocate a node to which the position of the active disk belongs, the calculation method for the data center thermally aware disk sequence includes the following steps:
t1, when the node to which the first active disk belongs is distributed, traversing all nodes, finding out the distributed node position with the lowest total data energy consumption in the corresponding energy consumption model, and recording the distributed node position;
t2, when the xth active disk is distributed, assuming that the previous x-1 disks are all fixed in the positions distributed in the previous step, searching the position with the lowest energy consumption of the data center corresponding to the xth disk, if the number of the disks distributed by the node is equal to the number of the disks owned by the node, searching the distribution again, otherwise, recording the position of the distributed node, wherein the value of x is 2 to the maximum number of the disks in the data center;
t3, sequentially increasing the number of active disks and allocating according to the step T2 until the maximum number of disks are allocated to the data center, and all recorded disk allocation node positions form a heat sensing disk sequence.
Step P3, as shown in fig. 3, the thermal sensing disk sequence is divided according to the set copy number and the active copy, and the specific division method is as follows:
Figure BDA0002284631290000101
wherein, divider is the dividing length of the disk number sequence, copyactiveCopy number of active diskinactiveFor inactive disk copy number, disknumFloor is rounded down for the total number of disks.
Step P4, for an incoming request, first storing in a request queue, and then according to the request type, performing allocation of the disk to which the copy belongs according to the thermally-aware disk sequence, where the request type is divided into a read request and a write request, the read request requires the disk where the active copy is located to be online, and for the write request, if a data set already exists, the disks of all copies are required to be online, otherwise, the copies are allocated according to the following method:
s1, distributing active copy
S11, partitioning the active copies of the heat sensing disk sequence, sequentially inquiring whether the disk has a residual space for storing a data set from the head to the tail of the partitioned sequence, if the disk with the first space is inquired, successfully allocating, and recording the copy allocation by using a data allocation table;
s12, inquiring whether the disk has a residual space for storing a data set from the last found disk position of the partition to the tail of the partition, if the inquired disk with the space is the first disk, the allocation is successful, and recording the copy allocation by the data allocation table;
and S13, repeating the previous step until all the required active copies are distributed.
S2, distributing the inactive copy
S21, partitioning the inactive copies of the heat sensing disk sequence, sequentially inquiring whether the disk has a residual space for storing a data set from the head to the tail of the partitioned sequence, if the disk with the first space is inquired, successfully allocating, and recording the copy allocation by using a data allocation table;
s22, inquiring whether the disk has a residual space for storing a data set from the last found disk position of the partition to the tail of the partition, if the inquired disk with the space is the first disk, the allocation is successful, and recording the copy allocation by the data allocation table;
and S23, repeating the previous step until all the required active copies are distributed.
And step P5, calculating the distribution of the active disks of the assigned tasks in the current time period to obtain the required cold air supply temperature, and setting the corresponding cooling temperature for the air conditioning device, so that the machine inlet temperature of each node is not higher than the specified temperature warning value.
In summary, in the embodiment, the energy consumption model considering both storage energy consumption and cooling energy consumption is adopted to calculate the total energy consumption of the data center, so that the air flow organization can be fully utilized to improve the refrigeration efficiency, reduce the refrigeration energy consumption and reduce the influence on the increase of the storage energy consumption. Compared with random placement and Hadoop-like copy placement, the placement method of the thermal sensing magnetic disk sequence is more energy-saving in performance under the condition of different read-write ratios, and the energy consumption of a data center is reduced to a greater extent; and in the disk awakening and sleep test evaluation, a thermal sensing disk sequence placement method is adopted, so that the disk awakening and sleep test times are less, and the damage to the disk is reduced.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.

Claims (9)

1. The energy-saving copy placing method based on airflow organization and oriented to the cloud data center is applied to data center copy placing and is characterized by comprising the following steps:
constructing a relation between refrigeration energy consumption and disk energy consumption according to a thermal cycle model among data center nodes to obtain a data center total energy consumption model;
calculating a thermal sensing disk sequence of the data center according to the obtained total energy consumption model;
dividing the thermal sensing magnetic disk sequence according to the set copy number and the active copies;
firstly storing an incoming request in a request queue, and then distributing a disk to which a copy belongs according to a thermal sensing disk sequence according to the request type;
and calculating the distribution of the disks allocated with the requests in the current time period to obtain the refrigeration supply temperature of the data center, and adjusting the corresponding air conditioner cooling temperature.
2. The cloud data center-oriented energy-saving copy placement method based on the air flow organization according to claim 1, wherein the relationship between the refrigeration energy consumption and the disk energy consumption is constructed according to a thermal cycle model among data center nodes, and the process of obtaining the total energy consumption model of the data center is as follows:
dividing the data center into a plurality of nodes according to the physical placement position of the disk, and modeling the air flow organization of the data center by adopting fluid dynamics CFD (computational fluid dynamics) to obtain energy consumption models of a disk storage system and an air-conditioning refrigeration system, thereby calculating the total energy consumption of the data center; and meanwhile, obtaining a model of the relationship between the total energy consumption of the data center and the number of the active disks in each node according to the model.
3. The cloud data center-oriented energy-saving copy placement method based on airflow organization according to claim 2, wherein in the process of dividing the disks into a plurality of nodes according to the physical placement positions of the disks, the disks in the same placement position area and sharing the power supply are classified as the same node according to the locality of the physical placement positions and the power supply sharing state of the disks.
4. The cloud data center-oriented energy-saving copy placement method based on the airflow organization according to claim 2, wherein in the process of modeling the airflow organization of the data center by using the fluid dynamics CFD, the fluid dynamics CFD is simulated and the thermal current organization conduction coefficient between the nodes is calculated according to the divided region nodes to obtain a thermal cycle influence coefficient matrix.
5. The cloud data center-oriented energy-saving copy placement method based on the air flow organization as claimed in claim 2, wherein in the process of obtaining the model of the relationship between the total energy consumption of the data center and the number of the active disks in each node according to the model, the mathematical relationship between the storage cooling energy consumption and the storage energy consumption is calculated according to the obtained thermal cycle influence coefficient matrix, and then the total energy consumption of the data center is obtained, so that the model of the relationship between the total energy consumption of the data center and the number of the active disks in each node is obtained.
6. The cloud data center-oriented energy-saving copy placement method based on the airflow organization according to claim 1, wherein in the process of calculating the thermal sensing disk sequence of the data center according to the obtained total energy consumption model, assuming that no active disk exists in the nodes in the data center, the number of active disks is sequentially increased, and nodes to which the positions of the active disks belong are allocated, and the calculation method includes the following steps:
t1, when the node to which the first active disk belongs is distributed, traversing all nodes, finding out the distributed node position with the lowest total data energy consumption in the corresponding energy consumption model, and recording the distributed node position;
t2, when the xth active disk is distributed, assuming that the previous x-1 disks are all fixed in the distributed positions in the previous step, searching the position with the lowest energy consumption of the data center corresponding to the xth disk, if the number of the disks already distributed by the node is equal to the number of the disks owned by the node, searching the distribution again, otherwise, recording the position of the distributed node, wherein the value of x is 2 to the maximum value of the number of the disks of the data center;
and T3, sequentially increasing the number of active disks and distributing according to the step T2 until the maximum number of disks are distributed to the data center, wherein all recorded disk distribution node positions form a heat sensing disk sequence.
7. The cloud data center-oriented energy-saving copy placement method based on airflow organization according to claim 1, wherein the process of dividing the thermal-aware disk sequence according to the set copy number and the active copies is as follows:
Figure FDA0002284631280000031
wherein, divider is the dividing length of the disk number sequence, copyactiveCopy number of active diskinactiveFor inactive disk copy number, disknumFloor is rounded down for the total number of disks.
8. The cloud data center-oriented energy-saving copy placement method based on the airflow organization as claimed in claim 1, wherein the incoming requests are first stored in a request queue, and then according to the request types, in the process of allocating the disks to which the copies belong according to the thermally-aware disk sequence, the request types are divided into read requests and write requests, the read requests require the disks where active copies exist to be online, and for the write requests, if a data set already exists, the disks of all copies are required to be online, otherwise, the copies are allocated according to the following method:
the allocation for the active copies is as follows:
s11, partitioning the active copies of the heat sensing disk sequence, sequentially inquiring whether the disk has a residual space for storing a data set from the head to the tail of the partitioned sequence, if the disk with the first space is inquired, successfully allocating, and recording the copy allocation by using a data allocation table;
s12, inquiring whether the disk has a residual space for storing a data set from the last found disk position of the partition to the tail of the partition, if the inquired disk with the space is the first disk, the allocation is successful, and recording the copy allocation by the data allocation table;
s13, repeating the previous step until all the required active copies are distributed;
the allocation for inactive copies is as follows:
s21, partitioning the inactive copies of the heat sensing disk sequence, sequentially inquiring whether the disk has a residual space for storing a data set from the head to the tail of the partitioned sequence, if the disk with the first space is inquired, successfully allocating, and recording the copy allocation by using a data allocation table;
s22, inquiring whether the disk has a residual space for storing a data set from the last found disk position of the partition to the tail of the partition, if the inquired disk with the space is the first disk, the allocation is successful, and recording the copy allocation by the data allocation table;
and S23, repeating the previous step until all the required active copies are distributed.
9. The cloud data center-oriented energy-saving copy placement method based on the air flow organization as claimed in claim 1, wherein the data center total energy consumption model is modeled according to the air flow organization, and the following energy consumption minimization model is obtained:
Figure FDA0002284631280000041
Figure FDA0002284631280000042
Request＝(DataSet,RequestType)
Y＝f(DataTable,Request)
wherein the content of the first and second substances,
P＝prunY+pidleλ
Figure FDA0002284631280000043
Figure FDA0002284631280000044
the data center is assumed to have m nodes, Request represents a task Request, DataSet represents a data set number of the Request, RequestType represents a Request type and is divided into a READ type and a WRITE type, DataTable represents a storage position of a data set in the data center, Y represents the number of online disks of each node, and Y is used foriTo represent the number of active disks per node of the data center, and Y represents YiVector of composition, P is the computing device energy consumption vector, PiThe ith in the P vector represents the storage energy consumption of each node, COP is the coefficient of performance of the air-conditioning refrigeration equipment, and tsupCooling temperature, lambda, supplied to the air-conditioning unitiThe binary variable of 0 or 1 indicates whether the ith node is in an active state, 0 indicates no, 1 indicates yes, and lambda is lambdaiComposed vector, idle active nodes can consume pidleIs represented by prunThe temperature alarm value is t and represents the additionally increased energy consumption for starting each active disk in a single nodecriticalThe heat cycle matrix D represents the relationship between the coefficient of the heat mutual influence between the nodes and the power consumption of the nodes.
CN201911155244.6A 2019-11-22 2019-11-22 Copy placement method based on airflow organization and oriented to cloud data center Pending CN110941396A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911155244.6A CN110941396A (en) 2019-11-22 2019-11-22 Copy placement method based on airflow organization and oriented to cloud data center

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911155244.6A CN110941396A (en) 2019-11-22 2019-11-22 Copy placement method based on airflow organization and oriented to cloud data center

Publications (1)

Publication Number Publication Date
CN110941396A true CN110941396A (en) 2020-03-31

Family

ID=69908002

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911155244.6A Pending CN110941396A (en) 2019-11-22 2019-11-22 Copy placement method based on airflow organization and oriented to cloud data center

Country Status (1)

Country Link
CN (1) CN110941396A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111859703A (en) * 2020-07-30 2020-10-30 暨南大学 Data center energy-saving data copy placement method based on heat sensing

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103530317A (en) * 2013-09-12 2014-01-22 杭州电子科技大学 Energy consumption adaptive type replication managing method used in cloud storage system
CN105892947A (en) * 2016-03-31 2016-08-24 华中科技大学 SSD and HDD hybrid caching management method and system of energy-saving storage system
US10242022B1 (en) * 2016-08-10 2019-03-26 Veritas Technologies Llc Systems and methods for managing delayed allocation on clustered file systems
CN109871268A (en) * 2019-01-10 2019-06-11 暨南大学 A kind of energy-saving scheduling method based on air current composition at data-oriented center

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103530317A (en) * 2013-09-12 2014-01-22 杭州电子科技大学 Energy consumption adaptive type replication managing method used in cloud storage system
CN105892947A (en) * 2016-03-31 2016-08-24 华中科技大学 SSD and HDD hybrid caching management method and system of energy-saving storage system
US10242022B1 (en) * 2016-08-10 2019-03-26 Veritas Technologies Llc Systems and methods for managing delayed allocation on clustered file systems
CN109871268A (en) * 2019-01-10 2019-06-11 暨南大学 A kind of energy-saving scheduling method based on air current composition at data-oriented center

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
王岩;汪晋宽;: "云存储中动态副本放置机制研究" *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111859703A (en) * 2020-07-30 2020-10-30 暨南大学 Data center energy-saving data copy placement method based on heat sensing

Similar Documents

Publication Publication Date Title
CN110134514B (en) Extensible memory object storage system based on heterogeneous memory
US7716425B1 (en) Prefetching data in distributed storage systems
US8141095B2 (en) Recording medium storing data allocation control program, data allocation control device, data allocation control method, and multi-node storage-system
CN111427844B (en) Data migration system and method for file hierarchical storage
Lee et al. PFRF: An adaptive data replication algorithm based on star-topology data grids
CN106156283B (en) Heterogeneous Hadoop storage method based on data temperature and node performance
US7818499B2 (en) Methods and apparatuses for heat management in storage systems
CN108932150B (en) Caching method, device and medium based on SSD and disk hybrid storage
JPH05225066A (en) Method for controlling priority-ordered cache
CN104699424A (en) Page hot degree based heterogeneous memory management method
US10649519B2 (en) Computer system, method for conserving power, and computer
Liao et al. Performing initiative data prefetching in distributed file systems for cloud computing
CN111367469A (en) Layered storage data migration method and system
Xie et al. Dynamic data reallocation in hybrid disk arrays
CN110795363A (en) Hot page prediction method and page scheduling method for storage medium
Wan et al. SSD-optimized workload placement with adaptive learning and classification in HPC environments
He et al. PSA: A performance and space-aware data layout scheme for hybrid parallel file systems
CN110941396A (en) Copy placement method based on airflow organization and oriented to cloud data center
CN104765572B (en) The virtual storage server system and its dispatching method of a kind of energy-conservation
JP5692355B2 (en) Computer system, control system, control method and control program
He et al. Cost-aware region-level data placement in multi-tiered parallel I/O systems
CN116737067A (en) Storage loading structure and method of graph data
CN109491985B (en) Association rule based thermal data processing method and apparatus
Lipetz et al. Automated tiering in a QoS environment using coarse data
Chen et al. Using working set reorganization to manage storage systems with hard and solid state disks

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20200331