WO2013068662A1 - Procédé, programme d'ordinateur et dispositif d'allocation de ressources informatiques d'un cluster pour l'exécution d'un travail soumis audit cluster - Google Patents
Procédé, programme d'ordinateur et dispositif d'allocation de ressources informatiques d'un cluster pour l'exécution d'un travail soumis audit cluster Download PDFInfo
- Publication number
- WO2013068662A1 WO2013068662A1 PCT/FR2012/052342 FR2012052342W WO2013068662A1 WO 2013068662 A1 WO2013068662 A1 WO 2013068662A1 FR 2012052342 W FR2012052342 W FR 2012052342W WO 2013068662 A1 WO2013068662 A1 WO 2013068662A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- cluster
- job
- computing resources
- work
- nodes
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5094—Allocation of resources, e.g. of the central processing unit [CPU] where the allocation takes into account power or heat criteria
-
- 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
- Y02B—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO BUILDINGS, e.g. HOUSING, HOUSE APPLIANCES OR RELATED END-USER APPLICATIONS
- Y02B70/00—Technologies for an efficient end-user side electric power management and consumption
- Y02B70/10—Technologies improving the efficiency by using switched-mode power supplies [SMPS], i.e. efficient power electronics conversion e.g. power factor correction or reduction of losses in power supplies or efficient standby modes
-
- 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
Definitions
- the present invention relates to the placement of jobs submitted to a cluster and more particularly to a method, a computer program as well as a computing resource allocation device of the cluster for the execution of jobs submitted to this cluster.
- Clusters generally include servers, data storage units, microprocessors and telematics equipment in the form of rows of computer cabinets ("racks" in English) in which computer equipment is installed, for example removably .
- a plurality of clusters is often brought together to form computer hosting centers known in the English terminology ("data center” or “server cluster”).
- Data centers typically have hundreds, sometimes even thousands, of electronic devices to interconnect via computer equipment.
- Each node therefore forms an independent computer and has its own characteristics such as, for example, the number of cores it comprises and a predetermined memory size (each core forming a computing unit).
- One or more nodes form a processing area, also called calculation area, and correspond to computer resources of a cluster.
- the processing zone may in certain cases be formed by only certain electronic components of a node, or even on the contrary by a large number of nodes.
- These clusters furthermore generally comprise a control and command unit equipped with a job management system, also called a job manager.
- the job manager is configured to receive a number of jobs ordered from outside the cluster.
- Processes for allocating computer resources in a cluster are known to execute jobs that are submitted to this cluster, which present the step of receiving the jobs and putting them in a queue. When a new job is submitted, it is added directly to the queue, for example at the end of the queue.
- the submitted jobs require computer resources corresponding to material needs, for example the number of cores needed, the required memory size and / or the priority level of execution of this work (for example minimal, normal or maximum).
- These known methods furthermore include the step of determining the allocation of the computer resources to the jobs listed in the queue and ready to be executed, according to the material characteristics of these works (material needs and priority) and the material and technical characteristics. availability specific to the treatment area (ie specific for example at each node of this area).
- the job manager then sends a selected job in the waiting list to the allocated computing resources so that the job is executed by the cluster.
- Each new job added to the job queue is thus executed according to the criteria of necessary available computer resources and priority.
- the invention thus provides, in a first aspect, a method for allocating computing resources of a cluster for the execution of at least one work submitted to said cluster, comprising a step of determining the placement of said at least one work submitted from material characteristics of said at least one job and hardware characteristics and availability of said computing resources at least one processing area of said cluster, said method being characterized in that said step of determining the placement of said at least one submitted work involves the following steps:
- the classification of the submitted and listed work is done according to a recommended list of allocation of computing resources for the execution of the work determined prior to their execution.
- This recommended list of computer resource allocation is advantageously determined on the basis of material data and availability data associated with the cluster processing area where Submitted and listed jobs could be executed, based on the material data associated with each work submitted and listed and based on energy data associated with that treatment area.
- this energetic aspect within the cluster makes it possible to precisely and finely place the submitted and listed work not based solely on the priority of a given work, the material needs of this given work as well as the availability of computing resources in the treatment area, but also based on the energy status of at least the treatment area and the energy impact of the work submitted and listed and ready to be performed at least on this treatment area, even on the whole cluster.
- the method according to the invention thus offers the possibility of correlating information on the energy state of at least the processing area, information on the hardware state and availability of computing resources at least in this area and the associated hardware information. to each of the works submitted and listed in order to place them optimally to execute them, in a simple, convenient and economical way.
- the method according to the invention makes it possible on the one hand to have a global vision of both the energy state and the material state at least of the treatment zone, or even of the entire cluster, and of the material status of the work submitted and listed, and on the other hand to correlate this information to determine the energy impact on at least the treatment area, or even the entire cluster, of the execution of the work by resources the treatment area.
- a global vision of both the energy state and the material state at least of the treatment zone, or even of the entire cluster, and of the material status of the work submitted and listed and on the other hand to correlate this information to determine the energy impact on at least the treatment area, or even the entire cluster, of the execution of the work by resources the treatment area.
- said energy state characteristics correspond to the heat dissipation and / or the electrical consumption of said computing resources of said cluster; and said method includes the step of determining heat dissipation characteristics of at least said computing resources of said processing area and / or the step of determining at least one of said computing resources of said treatment area;
- said hardware characteristics of said computing resources at least of said processing area of said cluster correspond to the number of basic computing units and / or to a memory size of said computing resources; and the method includes the step of determining said hardware characteristics of said computing resources and the step of determining said availability characteristics of said computing resources; said material characteristics of said at least one job correspond to the number of calculation base units and / or a memory size necessary for the execution of said at least one job and / or to an execution priority of said at least one job ; and the method comprises the step of determining a list of said hardware characteristics of said at least one job according to said number of calculation base units and / or said memory size necessary for the execution of said at least one job and / or said execution priority of said at least one job; and or
- it comprises the step of receiving an energy profile associated with said at least one job and the step of determining a list of said hardware characteristics of said at least one job according to said energy profile associated with said at least one job.
- the invention also provides, in a second aspect, a computer program comprising instructions configured to implement each of the steps of the method as described above when said program is executed on a computer.
- the invention also provides, in a third aspect, a device comprising systemic elements configured to implement each of the steps of the method.
- This device advantageously makes it possible to implement each of the steps of the method as described above in a simple, convenient and economical manner.
- said systemic elements are formed by a work manager and an energy manager each communicating independently with computing resources of a cluster; and a correlation system communicating with both said job manager and said energy manager.
- FIG. 1 very schematically represents a cluster provided with at least one processing zone and a device configured for implementing a method for allocating cluster computing resources for executing a submitted job. to this cluster, in accordance with the invention
- FIG. 2 very schematically represents a communication environment between one or more processing zones and the cluster device visible in FIG. 1;
- FIG. 3 illustrates an exemplary architecture of at least one computer resource of the cluster illustrated in FIG. 1;
- FIG. 4 is a block diagram illustrating various operating steps of the method
- FIGS. 5 and 6 are block diagrams respectively illustrating the step of establishing a base of energy state characteristics of at least one treatment zone and the step of establishing a hardware characteristics file of at least one less a submitted work as well as a basis of material characteristics and availability of said at least one zone, for the implementation of the operating steps visible in Figure 4;
- FIG. 7 is a block diagram showing, respectively, the step of establishing a heat dissipation characteristics file of each node forming a computing resource that comprises said at least one processing zone, the step of establishing a file. of the electrical consumption characteristics of each of these nodes, the step of establishing a hardware characteristics file of each of these nodes and the step of establishing an availability characteristics file of each of these nodes, for the implementation of steps visible in FIGS. 5 and 6; and
- FIGS. 8 and 9 are block diagrams illustrating other operating steps of the method.
- FIG. 1 very schematically illustrates a cluster 1 having a plurality of treatment zones 2, here six in number, as well as a general control and control device 6.
- Each treatment zone 2 forms a calculation zone which comprises several rows of computer cabinets 3 in which are inserted blades of computer equipment which are each formed of a plurality of electronic components.
- Each processing zone 2 has a plurality of nodes 5 corresponding to a basic computer equipment for the calculation, that is to say an independent computer.
- Each node 5 here forms a computing resource of cluster 1.
- Each node 5 has its own characteristics which will be described hereinafter with reference to FIG.
- the general control and control device 6 is provided with communication interfaces that make it possible, in particular, to receive input data coming from outside the cluster 1, such as, for example, the reception of T jobs to be executed within the cluster 1 .
- This general control and control device 6 is further provided with system elements 7, 8 and 9 which are respectively formed by a job manager 7, an energy manager 8 and a correlation system 9 which communicate and interact between them and with the nodes 5 of the cluster 1, as is described hereinafter with reference to FIG.
- FIG. 2 represents the interactive environment communicating between a processing zone 2 of the cluster 1, in particular a plurality of nodes 5 arranged in blades 4, with the systemic elements which are formed by the job manager 7, the manager of energy 8 and the correlation system 9.
- the treatment zone 2 is formed of four computer cabinets 3 each provided with a plurality of blades 4 arranged on three location heights which will be called high level, central level and low level with reference to FIG.
- Each blade 4 is provided with a plurality of nodes 5 whose number is here predetermined.
- the first computer cabinet 3 is provided with nodes 5 numbered from 0 to n
- the second computer cabinet 3 is provided with nodes 5 numbered from n + 1 to 2n + 1
- the third computer cabinet 3 is provided nodes 5 numbered 2n + 2 to 3n + 3
- the fourth computer cabinet 3 is provided with nodes 5 numbered 3n + 4 to 4n + 5.
- the job manager 7 is configured to receive input data submitted to the cluster 1, corresponding to data characteristic of a job T to be executed in at least the predetermined processing area 2 of the cluster 1.
- This job manager 7 is further configured to communicate reciprocally with both the nodes 5 and at least of the processing area 2 of the cluster 1 and with the correlation system 9.
- the job manager 7 is configured to receive information relating to the nodes 5 and coming from at least the processing zone 2, to transmit information to these nodes 5, to receive information from the correlation system 9 and to transmit information to this correlation system 9.
- the energy manager 8 is itself configured to communicate with both the nodes 5 at least of the processing zone 2 of the cluster 1 and with the correlation system 9; in that the energy manager 8 is configured to receive information relating to at least nodes 5 of this processing zone 2 and to transmit information to the correlation system 9.
- the correlation system 9 is configured to communicate with the energy manager 8 in that it is configured to receive information from this energy system 8; and to communicate reciprocally with the job manager 7 in that it is configured to receive information from this job manager 7 and to transmit information to this job manager 7.
- the processing zone 2 is configured to transmit information relating to the hardware and availability characteristics of the nodes 5 (forming the computing resources) to the job manager 7 and information relating to characteristics of the energy state of these nodes 5 to the energy manager 8.
- the energy manager 8 is itself configured to transmit an informative database relating to the energy characteristics at least from the processing zone 2 of the cluster 1 to the correlation system 9.
- the job manager 7 is itself configured to transmit information relating to the material characteristics of each work T that is submitted to the correlation system 9.
- the correlation system 9 is in turn configured to transmit information relating to a recommended allocation of the nodes 5 for the execution of at least one job T job manager 7, the latter being further configured to transmit information relating to at the nodes 5 (that is to say the computing resources) which have been chosen for the execution of this work T at least among the nodes 5 of the predetermined processing zone 2 itself chosen for the execution of the work T.
- FIG. 1 An exemplary architecture of a node 5 of the cluster 1 is illustrated in FIG.
- Node 5 here comprises a communication bus 51 to which are connected central processing units or microprocessors 52 (or CPU, acronym for "Central Processing Unit” in English terminology), components of RAM 53 (RAM, acronym for “Random Access Memory” in English terminology) having registers adapted to record variables of the parameters created and modified during the execution of programs, communication interfaces 54 configured to transmit and receive data; and internal storage elements 55, such as hard disks, including the executable code programs allowing nodes 5 to implement a job T.
- CPU central processing units
- RAM Random Access Memory
- each random access memory component 53 is associated with a microprocessor 52 or is common to the electronic components of the node 5.
- the communication bus 51 enables communication and interoperability between the various electronic components that comprise the node 5 or that are connected to it.
- the representation of the bus 51 is not limiting, and in particular, the microprocessors 52 are able to communicate instructions to any electronic component of the node 5 directly or via another electronic component of the node 5.
- program or programs implemented can be loaded by one of the electronic storage or communication components of the node 5 before being executed.
- the microprocessors 52 control and direct the execution of the instructions or portions of software code or programs that can be implemented in the node 5.
- the program or programs that are stored in a non memory volatile, for example a hard disk are transferred into the RAM 53 which then contains the executable code of the program or programs implemented, as well as registers for storing the variables and parameters necessary for the implementation of these programs.
- Each node 5 thus has a number of electronic components 52 to 55 as well as specific hardware characteristics.
- microprocessors 52 that it comprises, to the number of cores each forming a computing unit and each having a predetermined memory size, the total RAM size, the hard disk memory size, the maximum frequency of use of the microprocessors 52, the total use of the memory bandwidth and the inputs / outputs, the temperature of the node 5 or the individual temperature of an electronic component of the node 5 and the electrical power consumed by the node 5 or the individual electrical power consumed by an electronic component of node 5.
- node 5 These characteristics specific to node 5 are for some predetermined (number of microprocessors, number of cores, total memory size, size of RAM, size of hard disk memory and maximum frequency of use of microprocessors) and other characteristics are dynamic, that is to say that they vary according to the use of the node 5 and they are determined at a given time (use of the memory bandwidth and inputs / outputs, temperature, electrical power consumed and frequency of use of the microprocessor).
- FIG. 3 corresponds to that of a node 5, that is to say of a computing resource, of a processing zone 2 of the cluster 1 but that it could be acting identically, or at least similarly, the architecture of part of the general control and control device 6 which is configured to implement an algorithm described below with reference to FIGS. 9.
- microprocessors control and direct the execution of instructions or portions of software code of the program or programs according to the invention for the implementation of the method according to the invention, described below.
- FIGS. 4 to 9 of the method for allocating computer resources, that is to say nodes 5 of a predetermined processing zone 2 of cluster 1 for the execution of at least one T work submitted to cluster 1.
- FIG. 4 is a block diagram of the steps enabling the correlation system 9 to determine the placement of at least one job T submitted to the job manager 7.
- the correlation system 9 receives in step 100 a file listing the material characteristics of at least one work T. It will be seen below with reference to FIG. 6 what these material characteristics are.
- the correlation system 9 also receives in step 101 a database grouping the hardware characteristics of the computing resources, that is to say the nodes 5, of a processing zone. 2 and the availability characteristics of the nodes 5 of this processing zone 2. Below, we will see what the material characteristics of the nodes 5 of this zone 2 correspond to, also referring to FIG. 6.
- the predetermined processing area 2 corresponds to several nodes 5 of the cluster 1.
- the correlation system 9 also receives in step 102 a database grouping the energy state characteristics of the nodes 5 of the predetermined processing zone 2. It will be seen hereafter which information corresponds to these energy state characteristics. reference to Figure 5.
- the file received by the correlation system 9 in step 100 can group the hardware characteristics of several jobs T submitted to cluster 1 and received by the job manager 7.
- the databases received by the correlation system 9 in steps 101 and 102 can group the material, availability and energy state characteristics, not only of the predetermined processing area, but also of other processing areas of cluster 1, or even the entire cluster 1.
- the correlation system 9 determines in step 103 a recommended placement of the work T or jobs T listed in the file received in step 100 according to predetermined rules previously loaded by means of a file provided with these rules. in the general control and control device 6 to be taken into account in the correlation system 9.
- This determination step 103 is performed by correlating the hardware characteristics of the work or jobs T and the material characteristics, the availability and the energy state of the nodes 5 at least of the predetermined processing zone 2.
- the predetermined correlation rules of the above-mentioned characteristics relating to the work T and the nodes 5 of at least the predetermined processing zone 2 and the placement of these works T take into account the energy impact on the nodes 5 of the treatment zone.
- step 103 of determining a recommended placement of at least one work T among the works T comprises the step of determining the energy impact related to the execution of this work T among the works T from the material characteristics.
- the correlation system 9 then deduces at this same step 103 a recommended list of allocations of the nodes 5 of the predetermined processing area 2 for the execution of at least one work T among the jobs T in the cluster 1.
- the system correlation 9 thus generates a file establishing a recommended list of allocations of the nodes 5 for the execution of the job or jobs T and transmits (or sends) to the step 104 this file to the job manager 7 for the execution of or works T by the nodes 5, recommended or not, of the predetermined treatment zone 2.
- FIG. 8 is a block diagram of the steps for transmitting the choice of allocations of the nodes 5 of the predetermined processing zone 2 by the job manager 7 to this same zone 2.
- the job manager 7 receives in step 140 the file provided with the recommended list of allocations of the nodes 5 of the predetermined processing area 2 for the execution of one or more works T.
- the job manager 7 determines in step 141 the final allocation of the nodes 5 of the predetermined processing area 2 or of another processing area of the cluster 1 for the execution of one or more works T.
- This step 141 of determination is carried out according to rules provided with predominant criteria so that a weight (in other words a priority or an importance) can be put for example on the execution priority given to the works T or on the rules predetermined correlations implemented by the correlation system 9.
- the rules of the job manager 7 make it possible to define whether a user of the cluster 1 favors the saving of energy or the speed of submission of the work or jobs T.
- the job manager 7 then transmits (or sends) to the step 142 the file corresponding to the definitive choice of allocations of the job (s) T to the computing resources, that is to the nodes 5, of the treatment area. predetermined 2 of cluster 1, or another processing area according to the choice made by the job manager 7.
- step 141 this is a definitive choice in step 141, but it may be that this choice is modified according to the information received by the job manager 7, for example in the case where the correlation system 9 transmits a new file listing the recommended benefits.
- the correlation system 9 receives information (steps 100, 101 and 102) permanently, that is to say dynamically, and therefore determines a recommended list of allocations also dynamically (step 103) and is therefore likely to send such files listing the recommended allocations (step 104) also dynamically.
- FIG. 9 is a block diagram of the steps enabling the execution of job T or jobs T by nodes 5 of cluster 1.
- the nodes 5 of cluster 1 that are allocated for the execution of a job T receive in step 150 the file listing the final choices of allocations of these nodes 5 for the execution of at least one job T.
- FIG. 6 shows a block diagram of the steps for determining the list of the hardware characteristics of the job or jobs T submitted to the job manager 7.
- step 120 the job manager 7 receives a file of material characteristics of the job or jobs T that are submitted to it.
- the job manager 7 also receives in step 121 a file presenting an energy profile of each of the jobs T that are submitted to it.
- each work T corresponds to the number of basic calculation units necessary for the execution of this work T, and / or to a memory size necessary for the execution of work T and / or to a priority of execution of this work T.
- the energy profile of a work T is in fact a preliminary analysis of the energy impact of this work T with a precise statement of the needs in terms in particular of energy power, in other words electrical power consumed by the node (s) 5 which could perform this work T during its execution, and / or in terms of the heat dissipation of the node or nodes 5 which could execute this work T during its execution, in other words the temperature rise of this or these nodes 5.
- An energy profile of a work T for example results in characteristics of frequency of use of the processors 52 of a node 5 and / or percentage of use of the memory bandwidth of this node 5 and / or inputs /exits.
- the job manager 7 determines in step 122 a file listing the job or jobs T submitted to it and for each work T, the material characteristics of this work T associated with the energy profile of the latter.
- the job manager determines a list of the hardware characteristics of the job or jobs T that are submitted to it according to the number of basic units of calculation and / or the memory size necessary for the execution. of this or these work T and / or priority execution of this or these work T and / or the energy profile of this or these work T.
- the job manager 7 then transmits (or sends) to the step 123 the file listing the hardware characteristics of at least one job T (submitted to this job manager 7) to the correlation system 9 (which receives it at the end of the job). step 100 visible Figure 4).
- FIG. 7 shows a block diagram of the steps enabling the determination of the hardware characteristics of the nodes 5 at least of the predetermined processing zone 2 of the cluster 1 and the availability characteristics of these nodes 5.
- Cluster 1 determines in step 134 the hardware characteristics of each node 5 that it comprises.
- these material characteristics correspond to the number of basic units of calculation and / or to a memory size of each node 5.
- Cluster 1 transmits (or sends) to step 135 a file generated in step 134 provided with the hardware characteristics of at least nodes 5 of the predetermined processing area 2 of cluster 1, or even of the entire cluster 1, at manager of works 7.
- the cluster 1 also determines in step 136 the availability characteristics of each node 5 at least in the processing zone 2 of the cluster 1, or even of the entire cluster 1.
- Cluster 1 further transmits (or sends) at step 137 a file of the availability characteristics of the nodes 5 at least of the predetermined processing zone 2 of cluster 1, or even of the whole of cluster 1, which file was previously generated in step 136. This file is transmitted by cluster 1 to job manager 7.
- FIG. 6 further shows a block diagram of the steps enabling the determination of the material and availability characteristics of the predetermined processing zone 2, or even of the entire cluster 1.
- the job manager 7 receives in step 124 a file grouping the files generated in steps 134 and 136 by the cluster 1, ie a file grouping the hardware and availability characteristics of each node 5 at least in the predetermined processing area. 2, or even the entire cluster 1.
- the job manager 7 determines in step 125, from the characteristics associated with the nodes 5 received with the file in step 124, the hardware and availability characteristics of at least the predetermined processing area 2, or even all Cluster 1
- the job manager 7 generates in this same step 125 a database grouping the hardware and availability characteristics of at least the predetermined processing area 2, or even of the entire cluster 1, and transmits (or sends) this database. in step 126 to the correlation system 9 (the latter receives this database at step 120 visible in Figure 4).
- FIG. 7 further shows the block diagrams of the steps enabling the determination of the heat dissipation characteristics and the determination of the electrical consumption characteristics of the nodes 5 of at least the predetermined processing zone 2, or even of the entire cluster 1.
- Cluster 1 determines in step 130 the heat dissipation characteristics of each of these nodes 5, particularly corresponding to the predetermined processing zone 2, or even to the entire cluster 1.
- the cluster 1 generates at this same step 130 a file of characteristics of dissipation of the nodes 5 at least of the predetermined processing zone 2, or even of the whole of the cluster 1, then transmits (or sends) this file in the step 131 towards the correlation system 9.
- the cluster 1 further determines in step 132 the power consumption characteristics of each of the nodes 5 of the cluster 1, in particular the nodes 5 of the predetermined processing zone 2, or even of the entire cluster 1.
- the cluster 1 also generates at this same step 132 a power consumption characteristics file of the nodes 5 at least of the predetermined processing area 2, or even of the whole of the cluster 1, then transmits (or sends) this file to the step 133 to the correlation system 9.
- FIG. 5 is a block diagram of the steps enabling the determination of the energy characteristics of the predetermined processing zone 2, or even of the entire cluster 1.
- the energy state characteristics correspond to the heat dissipation and / or the electrical consumption, in other words to the electrical power consumed, of the nodes 5 at least of the predetermined treatment zone 2, or even of the cluster 1.
- the correlation system 9 also receives at steps 110 and 111, respectively, a file of the heat dissipation characteristics of the nodes 5 at least of the predetermined processing area 2, or even of the entire cluster 1, and a power consumption characteristics file. at least nodes 5 of the predetermined processing zone 2, or even of the entire cluster 1.
- the correlation system 9 determines in step 112 the energy state characteristics at least of the predetermined processing zone 2, or even of the entire cluster 1, according to the characteristics of the files received in steps 110 and 111.
- the correlation system 9 also generates at this same step 112 a database grouping the energy state characteristics at least of the predetermined processing zone 2, or even of the entire cluster 1, then transmits (or sends) to step 113 this database to another part of this correlation system 9, which receives at step 102 (FIG. 4) this database on which this correlation system 9 is based among other things to determine the recommended placement of the work T at step 103.
- the method of allocating the nodes 5 of at least one predetermined processing area 2 of the cluster 1 for the execution of at least one work T submitted to this cluster 1, in particular to its job manager 7, advantageously allows for 'improve the energy impact of the entire cluster 1 by reducing in particular the electrical power consumed by this cluster 1, which is particularly important in the context of clusters and computer hosting centers which, with the previous allocation processes above mentioned, can consume extremely high electrical powers.
- the job manager 7 receives a work T having a number of material characteristics as well as an energy profile.
- the hardware features of this T job are defined by a need of two hundred and fifty six cores, with 4GB of memory per core as well as a maximum execution priority.
- the energy profile of this work T defines an energy requirement requiring that the microprocessors 52 of the nodes 5 which will have to carry out this work T turn at full speed, that is to say that they have a maximum frequency, and requiring the total utilization of memory bandwidth and I / O, which means maximum energy consumption per node 5.
- the predetermined treatment zone 2 is provided with four computer cabinets 3 each having fifty-four nodes numbered respectively No. 0 to No. 53 for the first cabinet, No. 54 to No. 107 for the second cabinet, No. 108 to No. 162 for the third cabinet and No. 163 to No. 217 for the fourth cabinet.
- nodes No. 54 to No. 162 are nodes having 4 GB per core and having sixteen cores per node, the other nodes do not have these hardware characteristics.
- Nodes # 54 to # 162 are also nodes that are available for performing the T job.
- the file listing the material characteristics of the work T and the database containing the material and availability characteristics of the processing zone 2 was thus determined.
- the instantaneous consumption of the cluster 1, in particular of the predetermined processing zone 2 is as follows:
- the first computer cabinet 3 which has nodes No. 0 to No. 53 has on its three levels (high, central and low) a maximum energy consumption;
- the second computer 3 with nodes No. 54 to No. 107 has on its high level and on its low level a minimum energy consumption and on its central level average energy consumption;
- the third computer cabinet 3 provided with nodes No. 108 to No. 162 has on its three levels an average energy consumption;
- the fourth computer cabinet 3 with nodes No. 163 to No. 217 has on its high level and on its central level an average energy consumption and on its low level a minimum energy consumption.
- the first computer cabinet 3 provided with nodes No. 0 to No. 53 has on its three levels a maximum heat dissipation;
- the second computer cabinet 3 provided with nodes No. 54 to No. 107 has on its high level and on its low level an average heat dissipation and on its central level a maximum heat dissipation;
- the third and fourth computer cabinets 3 provided respectively with nodes No. 108 to No. 162 and No. 163 to No. 217 each have on its three levels an average heat dissipation.
- This information relating to the energy state of the nodes 5 of the predetermined processing zone 2 makes it possible to establish the database of the energy characteristics of the predetermined treatment zone 2, or even of the entire cluster 1.
- the job classification policy T that is to say the predetermined rules that are loaded into the correlation system 9 is based on the correlation between the current energy consumption (in other words before the execution of the work T), the current energy dissipation and the risk of extending the heat dissipation zone, that is to say that the nodes 5 which are in average or minimum heat dissipation are impacted by adjacent nodes 5 being in maximum heat dissipation and thus causing an increase in the temperature of the nodes 5 which are in average or minimum heat dissipation and thus an increase in their own heat dissipation.
- the correlation system 9 will then determine a file of recommended allocations of the nodes 5 for the execution of the work T.
- nodes n ° 54 to n ° 162 are potentially nodes able to execute the work T and thus these nodes can potentially be allocated to this work T.
- nodes n ° 54 to n ° 107 are more exposed in terms of energy consumption (in other words in electrical consumption) than the nodes n ° 108 to n ° 162.
- the requirements are first nodes at maximum power (for example having a maximum microprocessor frequency) and the placement policy (that is to say the predetermined rules of the system correlation 9) is to limit the impact on nodes 5 having a normal or minimum heat dissipation, it is therefore not advisable to take the nodes No. 108 to No. 162 given their geographical position. Indeed, the increase of the temperature of the nodes n ° 108 to n ° 162 would have an impact directly on the adjacent nodes n ° 54 with n ° 107 and n ° 163 with n ° 217.
- the recommended selection of the nodes 5 for the execution of the work T is therefore more judiciously chosen among the nodes # 54 to # 107.
- nodes Nos. 54 to 71 and / or nodes Nos. 91 to 107 corresponding respectively to the high level and the low level of the second computer cabinet 3 rather than the nodes Nos. 72 to 90.
- the nodes n ° 54 to n ° 71 are at the top of the second computer cabinet 3 (high level), it will be admitted that one privileges their use because the heat is evacuated more easily by the top.
- the blades 4 of electronic components in the frame of the second computer cabinet 3, there are two levels and it will be assumed that it is preferred to take the blades located on the top of the chassis than those located on the bottom of the frame . As a variant, one could have preferred to take a contrary rule or privilege continuity in the node numbers 5.
- the correlation system 9 recommends allocating the nodes Nos. 54, 56, 58, 60, 64, 66, 68, 70 on the high level. the second computer cabinet 3, then the nodes No. 55, No. 57, No. 59, No. 61, No. 62, No. 63, No. 65, No. 67 following the same high level of the same cabinet 3 .
- the correlation system 9 recommends the allocation of the nodes n ° 54 to n ° 68 and n ° 70 for the execution of the work T.
- no weight is assigned to the execution priority of the work T and / or to the energy criterion, in other words to the energy impact of this work T.
- the predetermined treatment zone does not correspond to an area as seen in FIG. 1 but rather corresponds to a complete row of computer cabinets 3, or even to a single computer cabinet 3, or even to a part of a computer cabinet 3 or even a single blade 4 of electronic component, or even a single node 5 or even a single part of a node 5, or even to the entire cluster 1;
- a node can be formed only of certain electronic components of a computer equipment, or even contrary to several computer equipment;
- the predetermined and loaded rules in the correlation system 9 are different from the rules taken in the example described;
- each job T submitted to the job manager 7 is provided with an energy profile, or on the contrary only certain jobs are associated with a respective energy profile or no work has an energy profile;
- the architecture of the node illustrated in FIG. 3 is different, for example it has more or less random access memory, more or fewer microprocessors;
- the step of determining energy state characteristics of the at least nodes of the predetermined processing zone is based solely on the heat dissipation characteristics of these nodes or only on the power consumption characteristics of these nodes;
- the step of determining the energy state characteristics of the nodes at least in the predetermined treatment zone is not performed by the correlation system 9 but rather by the energy manager 8; and or
- the data that passes between the energy manager, the correlation system, the job manager and the processing areas are not integrated into files or databases, but rather transit more generally via a network, at least for some of them, the others may for example be stored in files or databases or stored in memory before transmission.
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Power Sources (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Multi Processors (AREA)
Abstract
Description
Claims
Priority Applications (7)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
IN3731CHN2014 IN2014CN03731A (fr) | 2011-11-08 | 2012-10-15 | |
BR112014010884A BR112014010884A2 (pt) | 2011-11-08 | 2012-10-15 | método, programa de computador e dispositivo de alocação de recursos de informática de um agrupamento para a execução de um trabalho submetido ao dito agrupamento |
CA2852367A CA2852367C (fr) | 2011-11-08 | 2012-10-15 | Procede, programme d'ordinateur et dispositif d'allocation de ressources informatiques d'un cluster pour l'execution d'un travail soumis audit cluster |
EP12781396.2A EP2776927A1 (fr) | 2011-11-08 | 2012-10-15 | Procédé, programme d'ordinateur et dispositif d'allocation de ressources informatiques d'un cluster pour l'exécution d'un travail soumis audit cluster |
US14/356,874 US9880887B2 (en) | 2011-11-08 | 2012-10-15 | Method, computer program and device for allocating computer resources of a cluster for executing a task submitted to said cluster |
JP2014540533A JP6297980B2 (ja) | 2011-11-08 | 2012-10-15 | クラスタに依頼されたタスクを実行するために前記クラスタのコンピュータ資源を割り当てるための方法、コンピュータプログラム、およびデバイス |
CN201280054550.0A CN103946802B (zh) | 2011-11-08 | 2012-10-15 | 分配集群的信息资源以执行提交给该集群的工作的方法、计算机程序及设备 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
FR1160173 | 2011-11-08 | ||
FR1160173A FR2982386B1 (fr) | 2011-11-08 | 2011-11-08 | Procede, programme d'ordinateur et dispositif d'allocation de ressources informatiques d'un cluster pour l'execution d'un travail soumis audit cluster |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2013068662A1 true WO2013068662A1 (fr) | 2013-05-16 |
Family
ID=47143177
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/FR2012/052342 WO2013068662A1 (fr) | 2011-11-08 | 2012-10-15 | Procédé, programme d'ordinateur et dispositif d'allocation de ressources informatiques d'un cluster pour l'exécution d'un travail soumis audit cluster |
Country Status (9)
Country | Link |
---|---|
US (1) | US9880887B2 (fr) |
EP (1) | EP2776927A1 (fr) |
JP (1) | JP6297980B2 (fr) |
CN (1) | CN103946802B (fr) |
BR (1) | BR112014010884A2 (fr) |
CA (1) | CA2852367C (fr) |
FR (1) | FR2982386B1 (fr) |
IN (1) | IN2014CN03731A (fr) |
WO (1) | WO2013068662A1 (fr) |
Families Citing this family (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10541936B1 (en) | 2015-04-06 | 2020-01-21 | EMC IP Holding Company LLC | Method and system for distributed analysis |
US10331380B1 (en) | 2015-04-06 | 2019-06-25 | EMC IP Holding Company LLC | Scalable distributed in-memory computation utilizing batch mode extensions |
US10404787B1 (en) | 2015-04-06 | 2019-09-03 | EMC IP Holding Company LLC | Scalable distributed data streaming computations across multiple data processing clusters |
US10528875B1 (en) | 2015-04-06 | 2020-01-07 | EMC IP Holding Company LLC | Methods and apparatus implementing data model for disease monitoring, characterization and investigation |
US10791063B1 (en) * | 2015-04-06 | 2020-09-29 | EMC IP Holding Company LLC | Scalable edge computing using devices with limited resources |
US10776404B2 (en) | 2015-04-06 | 2020-09-15 | EMC IP Holding Company LLC | Scalable distributed computations utilizing multiple distinct computational frameworks |
US10812341B1 (en) | 2015-04-06 | 2020-10-20 | EMC IP Holding Company LLC | Scalable recursive computation across distributed data processing nodes |
US10496926B2 (en) | 2015-04-06 | 2019-12-03 | EMC IP Holding Company LLC | Analytics platform for scalable distributed computations |
US10509684B2 (en) | 2015-04-06 | 2019-12-17 | EMC IP Holding Company LLC | Blockchain integration for scalable distributed computations |
US10425350B1 (en) | 2015-04-06 | 2019-09-24 | EMC IP Holding Company LLC | Distributed catalog service for data processing platform |
US10366111B1 (en) | 2015-04-06 | 2019-07-30 | EMC IP Holding Company LLC | Scalable distributed computations utilizing multiple distinct computational frameworks |
US10706970B1 (en) | 2015-04-06 | 2020-07-07 | EMC IP Holding Company LLC | Distributed data analytics |
US10511659B1 (en) | 2015-04-06 | 2019-12-17 | EMC IP Holding Company LLC | Global benchmarking and statistical analysis at scale |
US10515097B2 (en) | 2015-04-06 | 2019-12-24 | EMC IP Holding Company LLC | Analytics platform for scalable distributed computations |
US10860622B1 (en) | 2015-04-06 | 2020-12-08 | EMC IP Holding Company LLC | Scalable recursive computation for pattern identification across distributed data processing nodes |
US10348810B1 (en) | 2015-04-06 | 2019-07-09 | EMC IP Holding Company LLC | Scalable distributed computations utilizing multiple distinct clouds |
US10541938B1 (en) | 2015-04-06 | 2020-01-21 | EMC IP Holding Company LLC | Integration of distributed data processing platform with one or more distinct supporting platforms |
US10122806B1 (en) | 2015-04-06 | 2018-11-06 | EMC IP Holding Company LLC | Distributed analytics platform |
US10015106B1 (en) * | 2015-04-06 | 2018-07-03 | EMC IP Holding Company LLC | Multi-cluster distributed data processing platform |
US10505863B1 (en) | 2015-04-06 | 2019-12-10 | EMC IP Holding Company LLC | Multi-framework distributed computation |
US10656861B1 (en) | 2015-12-29 | 2020-05-19 | EMC IP Holding Company LLC | Scalable distributed in-memory computation |
US10263838B2 (en) | 2016-08-23 | 2019-04-16 | International Business Machines Corporation | Assigning resources to a workload that utilizes embedded computing entities |
US10374968B1 (en) | 2016-12-30 | 2019-08-06 | EMC IP Holding Company LLC | Data-driven automation mechanism for analytics workload distribution |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110010456A1 (en) * | 2009-07-08 | 2011-01-13 | Fujitsu Limited | Recording medium storing load-distribution program, load-distribution apparatus, and load-distribution method |
EP2278465A2 (fr) * | 2009-07-07 | 2011-01-26 | Fujitsu Limited | Appareil d'attribution de tâche et procédé d'attribution de tâche |
US20110213508A1 (en) * | 2010-02-26 | 2011-09-01 | International Business Machines Corporation | Optimizing power consumption by dynamic workload adjustment |
Family Cites Families (36)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS597470Y2 (ja) | 1980-05-07 | 1984-03-07 | 株式会社武藤構造力学研究所 | 機器支持装置 |
JPS5794997U (fr) | 1980-12-01 | 1982-06-11 | ||
JPH01175800A (ja) | 1987-12-29 | 1989-07-12 | Fujitsu Ltd | キャビネットラックの相互連結構造 |
JPH0323237U (fr) | 1989-07-19 | 1991-03-11 | ||
US6179489B1 (en) | 1997-04-04 | 2001-01-30 | Texas Instruments Incorporated | Devices, methods, systems and software products for coordination of computer main microprocessor and second microprocessor coupled thereto |
JP4353341B2 (ja) | 1999-09-03 | 2009-10-28 | 株式会社昭電 | 免震装置 |
JP4523124B2 (ja) * | 2000-07-14 | 2010-08-11 | 日立アプライアンス株式会社 | エネルギサービス事業システム |
US7171668B2 (en) * | 2001-12-17 | 2007-01-30 | International Business Machines Corporation | Automatic data interpretation and implementation using performance capacity management framework over many servers |
US6964539B2 (en) * | 2002-03-18 | 2005-11-15 | International Business Machines Corporation | Method for managing power consumption of multiple computer servers |
US7689708B1 (en) | 2002-10-28 | 2010-03-30 | Netapp, Inc. | Apparatus to flow control frames in a networked storage virtualization using multiple streaming protocols |
US7127625B2 (en) * | 2003-09-04 | 2006-10-24 | Hewlett-Packard Development Company, L.P. | Application management based on power consumption |
US7197433B2 (en) * | 2004-04-09 | 2007-03-27 | Hewlett-Packard Development Company, L.P. | Workload placement among data centers based on thermal efficiency |
JP4855655B2 (ja) * | 2004-06-15 | 2012-01-18 | 株式会社ソニー・コンピュータエンタテインメント | 処理管理装置、コンピュータ・システム、分散処理方法及びコンピュータプログラム |
JP4895266B2 (ja) * | 2005-12-28 | 2012-03-14 | 富士通株式会社 | 管理システム、管理プログラムおよび管理方法 |
US7549070B2 (en) * | 2006-06-30 | 2009-06-16 | Sun Microsystems, Inc. | Method and apparatus for generating a dynamic power-flux map for a set of computer systems |
US7551971B2 (en) | 2006-09-13 | 2009-06-23 | Sun Microsystems, Inc. | Operation ready transportable data center in a shipping container |
US8667500B1 (en) * | 2006-10-17 | 2014-03-04 | Vmware, Inc. | Use of dynamic entitlement and adaptive threshold for cluster process balancing |
US9218213B2 (en) * | 2006-10-31 | 2015-12-22 | International Business Machines Corporation | Dynamic placement of heterogeneous workloads |
US7856549B2 (en) * | 2007-01-24 | 2010-12-21 | Hewlett-Packard Development Company, L.P. | Regulating power consumption |
US8284205B2 (en) * | 2007-10-24 | 2012-10-09 | Apple Inc. | Methods and apparatuses for load balancing between multiple processing units |
US8627325B2 (en) * | 2008-01-03 | 2014-01-07 | Hewlett-Packard Development Company, L.P. | Scheduling memory usage of a workload |
US8001403B2 (en) * | 2008-03-14 | 2011-08-16 | Microsoft Corporation | Data center power management utilizing a power policy and a load factor |
US8010815B2 (en) * | 2008-05-01 | 2011-08-30 | International Business Machines Corporation | Computational device power-savings |
US8086544B2 (en) * | 2008-09-03 | 2011-12-27 | International Business Machines Corporation | Analysis of energy-related factors for selecting computational job locations |
US9047083B2 (en) * | 2008-09-15 | 2015-06-02 | Vmware, Inc. | Reducing power consumption in a server cluster |
DE102008056412A1 (de) * | 2008-11-07 | 2010-05-12 | BSH Bosch und Siemens Hausgeräte GmbH | Haushaltsgerät mit einer Luft-Trocknungsvorrichtung und/oder Flüssigkeits-Heizungseinrichtung sowie zugehöriges Verfahren |
US9519517B2 (en) * | 2009-02-13 | 2016-12-13 | Schneider Electtic It Corporation | Data center control |
US9135104B2 (en) * | 2009-04-01 | 2015-09-15 | Soluto Ltd | Identifying frustration events of users using a computer system |
US8276139B2 (en) * | 2009-09-30 | 2012-09-25 | International Business Machines Corporation | Provisioning virtual machine placement |
US20110087522A1 (en) * | 2009-10-08 | 2011-04-14 | International Business Machines Corporation | Method for deploying a probing environment for provisioned services to recommend optimal balance in service level agreement user experience and environmental metrics |
JP5368285B2 (ja) * | 2009-12-11 | 2013-12-18 | 株式会社日立製作所 | 計算機システム、計算機リソースの管理方法及びプログラム |
US8671413B2 (en) * | 2010-01-11 | 2014-03-11 | Qualcomm Incorporated | System and method of dynamic clock and voltage scaling for workload based power management of a wireless mobile device |
JP5515889B2 (ja) * | 2010-03-15 | 2014-06-11 | 日本電気株式会社 | 仮想マシンシステム、自動マイグレーション方法および自動マイグレーションプログラム |
US8627123B2 (en) * | 2010-03-25 | 2014-01-07 | Microsoft Corporation | Managing power provisioning in distributed computing |
US8887169B2 (en) * | 2010-09-15 | 2014-11-11 | Empire Technology Development Llc | Task assignment in cloud computing environment |
US9417919B2 (en) * | 2012-09-06 | 2016-08-16 | Hewlett Packard Enterprise Development Lp | Computer cluster with objective-based resource sharing |
-
2011
- 2011-11-08 FR FR1160173A patent/FR2982386B1/fr active Active
-
2012
- 2012-10-15 WO PCT/FR2012/052342 patent/WO2013068662A1/fr active Application Filing
- 2012-10-15 EP EP12781396.2A patent/EP2776927A1/fr not_active Ceased
- 2012-10-15 CA CA2852367A patent/CA2852367C/fr not_active Expired - Fee Related
- 2012-10-15 US US14/356,874 patent/US9880887B2/en active Active
- 2012-10-15 BR BR112014010884A patent/BR112014010884A2/pt not_active Application Discontinuation
- 2012-10-15 IN IN3731CHN2014 patent/IN2014CN03731A/en unknown
- 2012-10-15 JP JP2014540533A patent/JP6297980B2/ja not_active Expired - Fee Related
- 2012-10-15 CN CN201280054550.0A patent/CN103946802B/zh not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP2278465A2 (fr) * | 2009-07-07 | 2011-01-26 | Fujitsu Limited | Appareil d'attribution de tâche et procédé d'attribution de tâche |
US20110010456A1 (en) * | 2009-07-08 | 2011-01-13 | Fujitsu Limited | Recording medium storing load-distribution program, load-distribution apparatus, and load-distribution method |
US20110213508A1 (en) * | 2010-02-26 | 2011-09-01 | International Business Machines Corporation | Optimizing power consumption by dynamic workload adjustment |
Also Published As
Publication number | Publication date |
---|---|
CA2852367C (fr) | 2019-08-27 |
US20140310718A1 (en) | 2014-10-16 |
EP2776927A1 (fr) | 2014-09-17 |
JP6297980B2 (ja) | 2018-03-20 |
FR2982386A1 (fr) | 2013-05-10 |
JP2014532946A (ja) | 2014-12-08 |
CN103946802A (zh) | 2014-07-23 |
IN2014CN03731A (fr) | 2015-09-04 |
BR112014010884A2 (pt) | 2017-05-02 |
US9880887B2 (en) | 2018-01-30 |
FR2982386B1 (fr) | 2016-05-27 |
CN103946802B (zh) | 2018-04-20 |
CA2852367A1 (fr) | 2013-05-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CA2852367C (fr) | Procede, programme d'ordinateur et dispositif d'allocation de ressources informatiques d'un cluster pour l'execution d'un travail soumis audit cluster | |
US20190342380A1 (en) | Adaptive resource-governed services for performance-compliant distributed workloads | |
US10439870B2 (en) | Assessment and dynamic provisioning of computing resources for multi-tiered application | |
EP2137944B1 (fr) | Propagation à la demande d'information de routage dans un système informatique distribué | |
US20080109547A1 (en) | Method, system and program product for determining a number of concurrent users accessing a system | |
US10726027B2 (en) | Cognitive elasticity of cloud applications | |
US20090113056A1 (en) | Computer resource distribution method based on prediciton | |
US20050108703A1 (en) | Proactive policy-driven service provisioning framework | |
US9176789B2 (en) | Capacity control | |
US10142404B2 (en) | Inter-platform management of computing resources | |
WO2014132009A1 (fr) | Procede de detection d'attaques de machines virtuelles | |
US20180205666A1 (en) | Application workload prediction | |
US10148531B1 (en) | Partitioned performance: adaptive predicted impact | |
US10142195B1 (en) | Partitioned performance tracking core resource consumption independently | |
US11016813B2 (en) | Optimizing initiator allocation | |
US20230244687A1 (en) | Optimization of Virtual Warehouse Computing Resource Allocation | |
US7925755B2 (en) | Peer to peer resource negotiation and coordination to satisfy a service level objective | |
Gohad et al. | Model driven provisioning in multi-tenant clouds | |
CN114730413A (zh) | 估计企业中的配置更改的结果 | |
Fazul et al. | Automation and prioritization of replica balancing in hdfs | |
WO2017055708A1 (fr) | Procede de controle de la capacite d'utilisation d'un systeme partitionne de traitement de donnees | |
US10397312B2 (en) | Automated server deployment platform | |
FR3031200A1 (fr) | Procede de gestion automatique de la consommation electrique d'une grappe de serveurs | |
EP3674897A1 (fr) | Procédé et dispositif d'aide à la décision pour l'allocation de moyens de calcul sur une infrastructure de calculs hautes performances | |
WO2019129958A1 (fr) | Procede de stockage de donnees et procede d'execution d'application avec reduction du temps d'acces aux donnees stockees |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 12781396 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2852367 Country of ref document: CA |
|
ENP | Entry into the national phase |
Ref document number: 2014540533 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 14356874 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2012781396 Country of ref document: EP |
|
REG | Reference to national code |
Ref country code: BR Ref legal event code: B01A Ref document number: 112014010884 Country of ref document: BR |
|
ENP | Entry into the national phase |
Ref document number: 112014010884 Country of ref document: BR Kind code of ref document: A2 Effective date: 20140506 |