FR3060791A1 - METHOD AND DEVICE FOR UPDATING - Google Patents

Abstract

The invention relates to a method of updating for at least one server of a cluster of a current value of a variable from an estimate of an already used quantity of a computing resource on the server or a maximum capacity of the server. server for this resource, this current value being taken into account by a cluster management device to select on receiving a request to execute a computer task, a target server to perform this task, the update being performed at each detection (F10) of an event (EV) out of a predefined set of events and: - if the detected event is to obtain a measurement of a current actual use of the resource on the server, the update (F30) uses the measure of the actual current usage of the resource on the server; if the detected event is the reservation for a computer task of a quantity of the resource on the server, the update (F40) uses the quantity of reserved resource; and if the detected event is the release by a computer task of a quantity of the resource on the server, the update (F50) uses the amount of resource released.

Description

® FRENCH REPUBLIC

NATIONAL INSTITUTE OF INDUSTRIAL PROPERTY © Publication number: 3,060,791 (to be used only for reproduction orders)

©) National registration number: 16 62694

COURBEVOIE © Int Cl ⁸ : G 06 F15 / 16 (2017.01), G 06 F 9/50

A1 PATENT APPLICATION

©) Date of filing: 16.12.16. ©) Priority: ©) Applicant (s): ORANGE Société anonyme - FR. @ Inventor (s): BROWN PATRICK and HA SON HAI. ©) Date of public availability of the request: 06.22.18 Bulletin 18/25. ©) List of documents cited in the preliminary search report: See the end of this booklet (© References to other related national documents: ©) Holder (s): ORANGE Société anonyme. ©) Extension request (s): ® Agent (s): CABINET BEAU DE LOMENIE.

FR 3 060 791 - A1 (04) METHOD AND DEVICE FOR UPDATING.

The invention relates to a method for updating for at least one server of a cluster with a current value of a variable from an estimate of an already used quantity of an IT resource on the server or a capacity. maximum of the server for this resource, this current value being taken into account by a cluster management device to select on receipt of a request to execute an IT task, a target server to execute this task, the update being performed on each detection (F10) of an event (EV) from a predefined set of events and:

- if the detected event is a measurement of a current actual use of the resource on the server, the update (F30) uses the measurement of the current actual use of the resource on the server;

- if the event detected is the reservation for a computer task of a quantity of the resource on the server, the update (F40) uses the quantity of resource reserved; and

- if the event detected is the release by a computer task of a quantity of the resource on the server, the update (F50) uses the quantity of resource released.

Invention background

The invention relates to the general field of computer systems, and more particularly computer systems based on clusters of machines (also more commonly referred to as "servers").

In a known manner, a cluster (or cluster) of servers groups together several independent computers (or nodes) placed in a network, so as to allow global and optimized management of computer resources (eg memory, CPU (Central Processing Unit), disk space. , network access, etc.) offered by these computers. Such a cluster of servers typically allows the distribution and launch in parallel, on several servers separate from the cluster, of different computer tasks associated with the same program or the same computer application (also designated by "work" or even "job" in English). In the context of current data processing, which can involve very large volumes of data (eg so-called “Big Data” processing), it is not uncommon to encounter clusters of servers comprising thousands or even tens of thousands of machines that need to be administered efficiently.

The distribution of tasks on the different machines of a cluster is generally administered by a cluster management system. Known cluster management systems are for example the Borg system offered by Google or the Hadoop Yarn system offered by The Apache Software Foundation. These systems are capable of administering clusters grouping together a very large number of machines, and of launching in parallel on these machines hundreds of thousands of tasks belonging to several thousands of different software applications.

In this context, cluster management systems are very often confronted with tasks which must be launched in bursts on the clusters. It is therefore necessary for them to efficiently process these tasks to make a decision very quickly as to their distribution on the various machines of the clusters. To this end, the prior art cluster management systems allocate the resources of a cluster to the different tasks of a program (in other words, associate with these different tasks different servers of the cluster capable of supplying the necessary resources. on their execution and reserve the required resources on these servers) on the basis of the resources reserved by the user of the cluster who requests the execution of his program.

Thus by way of example, the Hadoop Yarn cluster management system defines, for each server S of a Hadoop cluster and each IT resource R offered by the server, a limit capacity CL_R. For the sake of simplification, it is considered here that all the servers S have the same limit capacity for the resource R. The computer resource R can be of diverse nature such as for example memory, CPU (Central Processing Unit), space disk, network access, etc. To decide which resource to allocate on which server of the cluster to a task j of a program (or job), the Hadoop Yarn cluster management system quickly determines which servers are capable of executing this task j, in other words, which are servers which have the resources necessary for its execution in view of the R resource requirements declared by the program user. To this end, it uses the limit capacity CL_R of each server, the needs R_RJ of task j in resource R declared by the user of the program, and the total amount T_R of resource R already reserved on each server. A server is then considered capable of executing a task j only if for this server:

R_R_j + T_R <CL_R.

A server is then selected by the Hadoop Yarn management system from among the servers able to execute task j, according to predetermined criteria, not detailed here.

This mode of operation is not without constraints for users of the cluster.

Indeed, this requires users to specify the resource requirements of the different tasks of their programs in a sufficiently precise manner so as not to unnecessarily block resources at the cluster level nor, on the contrary, risk seeing their tasks not being processed or being processed less efficiently because they require more resources than those declared by the user. Such an estimation of the needs of the tasks of their programs is particularly difficult for users, especially since these needs are likely to evolve over time. In this context, it is classic for users to overestimate the resource requirements of the tasks of their programs, which leads to an under-utilization of the resources of the cluster.

Subject and summary of the invention

The invention makes it possible in particular to overcome the aforementioned drawbacks by proposing an updating method, for at least one server of a cluster of servers, of at least one current value of a variable denoted V chosen from among an estimate of a quantity already used of an IT resource on the server or a maximum capacity of the server for said IT resource, said at least one current value being intended to be taken into account by a server cluster management device to select, on receipt of a request for the execution of an IT task associated with a predetermined quantity of said IT resource initially required for its execution, a so-called target server of the cluster for executing this IT task, said method being such that said update is carried out at each detection of an event from a predefined set of events and:

If the detected event is a measurement of a current actual use of the computer resource on the server, said updating of the current value of the variable V uses said measurement of the current actual use of the computer resource on the server;

- if the detected event is a reservation by the management device for a computer task of a quantity of the computer resource on the server, said update of the current value of the variable V uses said quantity of the reserved computer resource ; and - if the detected event is a release by a computer task of a quantity of the computer resource on the server, said updating of the current value of the variable V uses said quantity of the computer resource released.

The invention also relates to a method for managing computer tasks comprising, on reception by a device for managing a cluster of servers of a request for execution of a computer task, this computer task being associated with a predetermined quantity of '' an IT resource initially required for its execution:

A step of determining, by the cluster management device, at least one server of the cluster capable of executing the computer task, this step of determining using the predetermined quantity of the computer resource initially required for the execution of the task computing, a maximum capacity of the server for the computing resource and an estimate of a quantity already used of the computing resource on the server;

A step of selection, by the cluster management device, of a server to execute the IT task among said at least one server of the determined cluster.

This method is remarkable in that, for at least one server of the cluster, at least one current value of a variable denoted V chosen from the estimate of the quantity already used of the computer resource on the server and the maximum capacity of the server. for the IT resource is updated in accordance with an updating method according to the invention.

No limitation is attached to the nature of the computer resources of the servers of the cluster managed by the cluster management device and required for the execution of the computer tasks in the context of the invention. It can be any type of computer and / or network resources (i.e. hardware or software components) such as memory, CPU, network access, disk space, etc.

In addition, the invention is described with reference to a single resource; however the same method can be used to take into account several resources, by applying it to each resource individually. It is during the actual selection of the server on which to execute the task that the plurality of resources will be taken into account by choosing a server having each of the resources necessary for the execution of the task.

The invention therefore proposes a solution which makes it possible to take account of the current and real (i.e. effective) use of resources at the level of the servers of a cluster to manage the allocation of resources within this cluster.

This solution is advantageously based on the current resource allocation mechanism applied by the prior art cluster management systems, which makes it possible to ensure a rapid response from the management device, particularly when tasks are in bursts. must be launched by it on the cluster. Indeed, without a priori knowledge of the actual use of resources by an IT task, the management device reserves for this task at the level of the server on which it launches, a quantity of resource equal to the need declared by the user to the origin of the computer task (predetermined quantity of the computer resource initially required for the execution of the computer task within the meaning of the invention).

The state-of-the-art mechanism is however adapted so as to take into account, after the launch of a task by the management device on a server of the cluster, the actual use of resources by this task on the server.

This offers the possibility to the management device, if this actual use is less than the needs of the task declared by its user, to allocate the resources not used by the task to other computer tasks presented subsequently to the management device. This is taken into account by updating, from measurements carried out directly on the servers, the quantities used by the management device to decide on the allocation of resources, namely the estimation of the quantity already used the IT resource required by the task on the server and / or the maximum server capacity for this resource. In a particular embodiment, obtaining the measurements of current use of the resource can be periodic.

In addition, in accordance with the invention, other events affecting the resources available on the server trigger an update of the current value of the variable V considered. These events are the reservation on the server, by the management device, of a quantity of the resource considered for other tasks, as well as the release of a quantity of this resource on the server by tasks which have finished run on it.

This makes it possible to take into account the computer tasks which are positioned asynchronously on the server by the management device between two feedbacks of measurements of real use of the resource on the server (these feedbacks are generally not continuous but periodic or at predetermined times), as well as those that end on the server. Taking into account completed tasks has a privileged interest when an IT task launched on a server ends very quickly, typically in a time less than the reporting period of the actual usage measurements made on the server if necessary. This eliminates the remaining resource reserved for this task on the server and which is no longer used by this server without being reflected in the measurement.

The invention therefore makes it possible to pass on in real time the various variations likely to affect the amount of resource available on the server in the amount taken into account by the management device to respond to a request for the execution of an IT task. . The cluster management device has a more precise representation of the amount of resource actually used on the server at all times, or at least when it receives a request to execute a task.

The cluster management device can thus simultaneously, thanks to the invention, learn how the resources are actually used on the servers while being able to make decisions quickly to allocate resources to the computer tasks presented to it. This rapid decision-making is carried out instantly on the basis of the resources which are reserved by the users for these tasks without waiting to know how these tasks will actually behave. The invention thus solves the double problem of addressing bursts of tasks without having knowledge of their future use of the resources of the cluster while allowing to allocate to other tasks unused resources of the cluster.

The invention therefore provides a solution compatible with large clusters and software applications comprising a large number of computer tasks to be executed on these clusters, and which makes it possible to gradually correct an erroneous or imprecise estimate of the resource needs of these tasks. by their users. The constraints imposed on cluster users can thus be relaxed.

Furthermore, the resources of the clusters are better used while trying to ensure the correct execution of the tasks on these clusters: in fact, thanks to the invention, the prolonged waiting times for the execution of a task are limited IT or crowding out an IT task, in particular in the event of insufficient resources allocated to this task, taking into account the actual use of resources on the cluster servers. It should be noted that a prolonged waiting time or crowding out a task is always possible, however he or she is offset by the fact that a greater number of tasks can be launched simultaneously on the cluster due to better management of the latter's resources. The performance of the cluster is thus improved overall.

In addition, the invention makes it possible to adapt to a possible variation over time of the resource requirements of the tasks launched on the servers of the clusters.

It should be noted that updating the maximum capacity of the server for the resource rather than estimating the quantity already used of the computer resource required by the task on the server advantageously makes it possible to use as a device for managing the cluster l 'one of the state of the art management systems. Indeed, the maximum capacity of each server is information which is transmitted periodically today by this server to the cluster management system in the state of the art. By updating this capacity at the server level to reflect the actual use of the resource, we ensure that the cluster management system takes into account this real use for the allocation of cluster resources in a way completely transparent to him. In other words, this makes it possible to envisage an implementation in which the functioning of the centralized management system of the cluster remains the same (and therefore in which it is possible to use state of the art management systems, such as the Borg or Hadoop Yarn systems mentioned above), and where only the servers in the cluster must be adapted to take into account in the maximum capacity that they go back to the cluster management system, the actual use of their resources. Such an implementation is particularly interesting because it allows a progressive installation of the invention on all of the servers of the cluster while keeping the cluster management system unchanged. Updating servers is in practice a much simpler task than updating the cluster management system, which in itself is often an already very complex module.

In addition, in this variant, the invention can rely on the signaling (i.e. communication protocol) already defined between the servers of the cluster and the management system of the cluster.

In a particular embodiment, if the event detected is the obtaining of the measurement of the current actual use of the computer resource on the server, the updating of the current value of the variable V also uses at least a value of the variable V denoted Vpast updated at a previous instant.

In this embodiment, during the update, the measurement of the current actual use of the computer resource can in particular be weighted by a real factor a between 0 and 1, and said at least one value Vpast be weighted by a factor equal to (Ια).

This embodiment, which implements a moving average of exponential type, makes it possible to smooth over time the taking into account of the real use at the server level of the resource relative to the quantity of resource initially reserved. This smoothing gives the possibility of a progressive taking into account of the real use of resources at the level of each server while avoiding sudden transient fluctuations which could lead to a less efficient allocation of resources by the cluster management device. In fact, the closer the factor a is chosen to 0, the longer one takes into account the quantity of resources reserved and the use of the resources available in the cluster by the management device is delayed. On the contrary, by choosing a factor a close to 1, the management system can use the resources available earlier in the time. It is noted that the factor a can vary over time.

Of course, other ways of taking into account the measurement of the actual use of the computer resource and, where appropriate, said at least one value updated at a previous instant can be envisaged. We can rely for example on the principle of an arithmetic or weighted moving average of measurements obtained at different times and evaluated on a sliding window of predetermined size whose end is positioned at the current instant, etc. In the case of a weighted average, we can consider applying decreasing weights as the age of the measurements increases.

By way of example, in an alternative embodiment of the invention in which the updated variable V is the estimate EU_R of the quantity already used of the computer resource on the server, if the event detected is the obtaining from the measurement of the current actual use of the computer resource on the server, said current value denoted EU_R (n) of the variable V is updated according to the equation:

EU_R (n) = max ((l - a) EU_R (n - 1) + aU_R (n), U_R (n)) where n denotes an integer greater than 1, EU_R (n - 1) is the value of the estimation of the quantity already used of the computer resource on the server updated at a previous instant, and U_R (ri) designates the measurement obtained from the current actual use of the computer resource on the server.

This variant has a privileged application when the IT resource in question is memory. Indeed, the constraints in terms of memory resource are relatively strong constraints in the sense that a task can be ousted from a machine on which it is executed if it requires at a given instant more memory than the machine has available for this purpose. instant. By forcing the update of the value EU_R (n) to a value greater than or equal to the memory resource U_R (n) actually used on the server when the value of (1 - a) EU_R (n - 1) + aU_R ( n) does not sufficiently reflect the actual occupation of the memory on the server, it is ensured that the management device does not unduly allocate new tasks to the memory on the server which could then be missed by the computer tasks s 'already running on the server and prove to be harmful for the processing of these.

In another embodiment of the invention, if the detected event is the obtaining of the measurement of the current actual use of the computer resource on the server, this measurement corresponds to a use of the computer resource on the server by one or more IT tasks running on the server and all managed by the cluster management device.

On a server can indeed be executed in addition to the computer tasks launched on this server by the cluster management device, other processes such as typically processes internal to the server launched for example by its operating system. In this embodiment, it is ensured that only the tasks managed and positioned properly speaking by the cluster management device are taken into account by the latter when allocating resources to the tasks presented to it. In other words, this embodiment allows a forceful intervention of the management device in relation to the rights (i.e. quantities of resource) reserved for it on the server.

One way of implementing this embodiment is for example as follows:

The measurement of the current actual use of the computer resource on the server comprises a plurality of values denoted U_R (n, j), j = l, ... J, associated respectively with a plurality of computer tasks Tj, j = l, ... J, positioned by the cluster management device on the server, J and n designating integers greater than or equal to 1, each value U_R (n, f) corresponding to a measure of the actual use of the IT resource on the server by the task Tj;

- said update (F30) includes:

a first evaluation step for each task Tj, of a current value EU_R (n, j) of an estimate of an amount already used of the computer resource on the server by the computer task Tj according to the equation:

EU_R (n, j) = max ((l - a) EU_R (n - l, y) + aU_R (n, j), U_R (n, j)) or according to the equation:

EU_R (n, f) = (1 - a) EU_R (n - IJ) + aU_R (n, f) or

EU_R (n - i ,; ') designates a value evaluated at a previous instant of the estimation of the quantity already used of the computer resource on the server by this computer task Tj; and a is a real factor between 0 and 1;

a second step of evaluation of the current value noted EU_R (n) of the estimation of the quantity already used of the computer resource on the server according to the equation:

j

EU_R (n) = ^ EU_R (n, j ') i = i or according to the equation:

EU_R (ri) = max

where J denotes the number of IT tasks positioned on the server by the cluster management device at the time of the second evaluation step.

In this variant, we therefore have access to more precise information on the real use of the server, since these are measures of the real use of the server for each computer task positioned on the server (i.e. 'actually running on it or for which a quantity of the resource has been reserved by the management device on the server) which are available and used when updating the value EU_R (n). In this way, when a task disappears and frees up the resource on the server, this is taken into account instantly in its contribution to the quantity EU_R (n) (the sum made to evaluate EU_R (n) only takes account of tasks positioned on the server at this time, that is to say at the time of the evaluation of EU_R (n)).

We note however that this variant requires additional calculations to update the quantity EU_R (n), linked to the individual management of tasks Tj, j = l, ..., J. These calculations can preferably be implemented locally at the level of each server and then be returned to the cluster management device by the nodes.

In another embodiment of the invention in which the updated variable V is the estimate EU_R of the quantity already used of the computer resource on the server, and if the event detected is the obtaining of the measurement of the current actual use of the computer resource on the server, said current value denoted EU_R (n) of the variable V is updated according to the equation:

EU_R (ri) = U_R (ri) + DR_R (ri) where U_R (n) denotes a current measure of the actual use of the computer resource on the server, and DR_R (n) denotes a current value of a quantity of the IT resource reserved for at least one IT task by the management device on the server.

In addition, in this embodiment, the method can further comprise:

A step for updating the current value DR_R (n) of the quantity of the IT resource reserved for at least one IT task by the management device on the server is obtained according to the equation:

DR_R (n) = (1 - P) DR_R (n - 1) where β denotes a real factor between 0 and 1 and DR_R (n - 1) denotes a quantity of the IT resource reserved by the management device on the server obtained at a previous instant;

- a step of detecting the update of the current value DR_R (n); and - a step of updating the current value of the variable EU_R according to the equation:

EU_R (n) = U_R (n) + DR_R (n).

In this embodiment, the weighting (by a factor (l - /?) Here) is applied no longer on the measurement of the actual use of the resources on the server but on the resources which are reserved by the device for managing the cluster on this server. In other words, the reserved resource quantities are smoothed and no longer the actual use of the resource. This embodiment ensures that the cluster management device takes into account, when allocating resources to a new task, a quantity of this resource used on the server always at least equal to its actual use (in other words, we in this embodiment eliminates the need to use a “max” function as in the examples described above).

In another embodiment, the updating method further comprises:

- if the detected event is a reservation by the management device for a computer task of a quantity R_R_Tnew of the computer resource on the server, a step of updating the current value DR_R (n) of the quantity of the IT resource reserved for at least one IT task by the management device on the server using said quantity R_R_Tnew and a quantity DR_R (n - 1) of the IT resource reserved by the management device on the server obtained at a previous instant, - if the detected event is a release by an IT task of an amount R_R_Told of the IT resource on the server, a step of updating the current value DR_R (n) of the amount of the IT resource reserved for at least a computer task by the management device on the server using said quantity R_R_Told and quantity DR_R (n - 1);

and in which said updating of the current value of the variable EU_R uses said quantity of the computer resource reserved according to the equation:

EU_R (n) = U_R (ri) + DR_R (n).

The choice of one or other of the embodiments described above may depend on different parameters, such as in particular the context in which the invention is implemented or even a compromise between complexity of implementation and performance of the allocation of resources by the management system.

In a particular embodiment, the updating method is implemented by the cluster management device.

Correlatively, in this embodiment, the invention also relates to a device for managing a cluster of servers comprising:

- a reception module;

A determination module, activated on reception by the reception module of a request to execute an IT task, said IT task being associated with a predetermined quantity of an IT resource initially required for its execution, the determination being configured to determine at least one server of the cluster able to execute the computer task using the predetermined quantity of the computer resource initially required for the execution of the computer task, a maximum capacity of the server for the computer resource and an estimate d 'an already used amount of IT resources on the server;

- a selection module configured to select a server to execute the computer task among said at least one server of the cluster determined by the determination module;

The management device according to the invention is remarkable in that it further comprises:

- a detection module configured to detect a predefined set of events; and - an update module activated for at least one server of the cluster, and configured to update at least one current value of a variable denoted V chosen from the estimate of the quantity already used of the computer resource on the server and the maximum capacity of the server for the IT resource, said update module being activated on each detection of an event from the predefined set of events by the detection module, and configured for:

o if the detected event is a measurement (U_R (n)) of a current actual use of the computer resource on the server, update the current value of the variable V using said measurement of the current actual use of the computer resource on the server;

o if the detected event is a reservation by the management device for a computer task of a quantity of the computer resource on the server, update the current value of the variable V using said quantity of the reserved computer resource; and o if the detected event is a release by a computer task of a quantity of the computer resource on the server, update the current value of the variable

V using said quantity of the computer resource released.

In another embodiment, the updating method is implemented by the server. The management method then further comprises a step of transmitting said at least one value updated by the server to the cluster management device.

Correlatively, in this embodiment, the invention also relates to a server belonging to a cluster of servers managed by a management device, said server being able to execute at least one computer task managed by said management device, said at least one computer task using at least one computer resource from the server during its execution, this server comprising modules activated for at least one time interval and comprising:

- a measurement module configured to measure an actual use of the IT resource on the server;

- a detection module configured to detect a predefined set of events; and an update module configured to update at least one current value of a variable denoted V chosen from an estimate of an amount already used of the computer resource on the server and a maximum capacity of the server for the resource. IT, said update module being activated on each detection by the event detection module from among the predefined set of events and configured for:

o if the detected event is a measurement of a current actual use of the computer resource on the server, update the current value of the variable V using said measure of the current actual use of the resource computer on the server;

o if the detected event is a reservation by the management device for a computer task of a quantity of the computer resource on the server, update the current value of the variable V using said quantity of the reserved computer resource; and o if the detected event is a release by a computer task of a quantity of the computer resource on the server, update the current value of the variable

V using said quantity of the computer resource released; and a transmission module, configured to transmit to the management device said current value of the updated variable V.

As mentioned previously, this embodiment can make it possible to facilitate the progressive implementation of the invention by limiting the developments necessary for its implementation carried out on the cluster management device.

In a particular embodiment, the different steps of the updating method and / or the different steps of the management method are determined by instructions of computer programs.

Consequently, the invention also relates to a computer program on an information medium, this program being capable of being implemented in a management device, in a server or more generally in a computer, this program comprising instructions adapted to the implementation of the steps of an updating process as described above.

The invention also relates to a computer program on an information medium, this program being capable of being implemented in a management device or more generally in a computer, this program comprising instructions adapted to the implementation steps of a management method as described above.

Each of these programs can use any programming language, and be in the form of source code, object code, or intermediate code between source code and object code, such as in a partially compiled form, or in any what other desirable form.

The invention also relates to an information or recording medium readable by a computer, and comprising instructions of a computer program as mentioned above.

The information or recording medium can be any entity or device capable of storing the program. For example, the support may include a storage means, such as a ROM, for example a CD ROM or a microelectronic circuit ROM, or else a magnetic recording means, for example a floppy disk or a disc. hard.

On the other hand, the information or recording medium can be a transmissible medium such as an electrical or optical signal, which can be routed via an electrical or optical cable, by radio or by other means. The program according to the invention can in particular be downloaded from a network of the Internet type.

Alternatively, the information or recording medium can be an integrated circuit in which the program is incorporated, the circuit being adapted to execute or to be used in the execution of the process in question.

The invention also relates to a computer system comprising:

A cluster of servers comprising a plurality of servers, each server comprising at least one IT resource allowing the execution of IT tasks;

- a server cluster management system;

wherein the cluster management device is in accordance with the invention and / or at least one server in the cluster is in accordance with the invention.

It is also conceivable, in other embodiments, that the updating method, the management method, the management device, the server and the computer system according to the invention have all or some of the above characteristics in combination. .

Brief description of the drawings

Other characteristics and advantages of the present invention will emerge from the description given below, with reference to the appended drawings which illustrate an embodiment thereof devoid of any limiting character. In the figures:

- Figure 1 shows, schematically, a computer system according to the invention, in a particular embodiment, the computer system comprising a cluster of servers and a cluster management device;

- Figure 2 shows the hardware architecture of the cluster management device of Figure 1;

- Figure 3A shows, in the form of a flowchart, the different steps of a management method as it is implemented in a particular embodiment by the management device of Figure 1;

FIG. 3B represents, in the form of a flowchart, the different steps of an updating method as it is implemented in a particular embodiment by the management device of FIG. 1; and FIG. 3C represents, in the form of a flowchart, the different steps of an updating method in an alternative embodiment.

Detailed description of the invention

FIG. 1 represents, in its environment, a computer system 1, in accordance with the invention, in a first embodiment.

According to the invention, the computer system 1 comprises - a cluster 2 of servers comprising a plurality of servers SI, S2, ..., SK, K designating an integer greater than 1, connected in network. Each server Sk, k = l, ..., K includes various computer resources (hardware and / or software) allowing the execution of computer tasks. These resources are for example memory, disk space, CPU, network access, etc. No limitation is attached to the nature of the IT resources offered by the servers SI, ..., SK of the cluster 2. However, to better illustrate the invention, we attach here to a particular IT resource R which consists of memory; and a device 3 for managing the cluster 2 of servers, in accordance with the invention.

The cluster management device 3 is configured to, on receipt of a request

REQ from a client device 4 (eg computer) of a user of cluster 2 requesting the execution of at least one IT task T, allocate resources of cluster 2 to this client for the execution of this task . More specifically, on reception of the REQ request, the device 3 for managing the cluster 2 identifies a server Sk of the cluster 2 capable of executing the computer task T, that is to say a server Sk having the resources necessary for its execution (memory, CPU, etc.), then launches the computer task T on this Sk server so that it is executed there.

No limitation is attached to the nature of the computer task T to be executed on cluster 2. It can be any type of task associated with an APP software application installed on the client device 4 or with any program. IT in general. In a known manner, such a computer program comprises a plurality of computer tasks to be executed, which can be subjected in bursts to the device 3 for managing the cluster 2 in order to be executed in parallel on different servers of the cluster 2. We are concerned here more particularly to the processing performed by the management device 3 on one of these tasks, namely on the task T. A similar processing is applied to all of the tasks contained in the request REQ.

Each of the tasks sent by the software application APP to the device 3 for managing the cluster 2 has IT resource requirements to enable its execution on the cluster 2. These needs are specified here by the user of the client device 4 for each of the tasks and for each of the resources. For the computer resource R and the task T, the user of the client device 4 specifies an amount R_R_T of memory required "initially" for the execution of the task T. The amount R_R_T can be roughly determined by the user of the client device 4, based on its knowledge of the operations implemented by the computer task T and the consumption in terms of memory required by these operations. It is noted that although the user of the client device 4 tries as far as possible to provide a quantity R_R_T close to the real use of the resource R by the computer task T, this quantity R_R_T is difficult in practice to estimate precisely for the user: it is generally overestimated by the user so as to ensure that the computer task T is allocated a sufficient amount of memory by the device 3 for managing the cluster 2 so that it can be executed in good conditions. The quantity R_R_T thus predetermined is transmitted here in the request REQ for execution of the computer task T addressed to the device 3 for managing the cluster 2.

In the embodiment described here, the device 3 for managing the cluster 2 has the hardware architecture of a computer, as illustrated in FIG. 2. It includes in particular a processor 5, a random access memory 6, a read only memory 7, a non-volatile flash memory 8 as well as communication means 9 enabling it to communicate in particular on the one hand with the client device 4 and on the other hand with the servers Sk, k = l, ..., K of the cluster 2. These communication means 9 comprise for example a network card, well known in itself.

The read only memory 7 of the device 3 for managing the cluster 2 constitutes a recording medium in accordance with the invention, readable by the processor 8 and on which is recorded here a computer program PROG in accordance with the invention.

The computer program PROG defines functional modules (and software here), configured to implement the steps of the updating method and the steps of the method of managing computer tasks according to the invention. These functional modules are based on and / or control the hardware elements 5-9 of the management device 3 mentioned above. They include in particular here, as illustrated in FIG. 1:

A module 3A for receiving requests from client devices such as the client device 4, this module 3A relying on the communication means 9;

A module 3B for determining, activated on reception by the module 3A for receiving a request to execute an IT task, the determining module being configured to determine at least one server of cluster 2 capable of executing the IT task ;

A selection module 3C configured to select a server to execute the computer task among the servers of the cluster determined by the determination module 3B;

- a 3D detection module, configured to detect a predefined set of events relating to the resource R and detailed later; and - an update module 3E, configured to update at each event detection by the 3D detection module, one or more values used by the determination module 3B to determine the servers capable of executing the computer task using the measurements of real use of the IT resource R obtained by the 3D obtaining module.

The functions of these different modules are now described in more detail, with reference to FIG. 3 (consisting of FIGS. 3A and 3B) and to the steps of the updating method and of the method for managing computer tasks according to the invention, such as that they are implemented, in the first particular embodiment, by the device 3 for managing the cluster 2.

As illustrated in FIG. 3, the method for managing computer tasks according to the invention is based, in the first embodiment described here, on two processes which take place in parallel:

An asynchronous processing PI process, illustrated in FIG. 3A, which is triggered by the device 3 for managing the cluster 2 upon reception by its module 3A for reception of each request for execution of computer task (s) on cluster 2. This PI process aims to allocate to each IT task computer resources on a server of cluster 2 in order to allow its execution; and - an update process P2, illustrated in FIG. 3B, which is implemented here continuously by the device 3 for managing the cluster 2. More particularly here, the updates target values used by the device 3 management of cluster 2 to select a server capable of executing an IT task entrusted to it. These updates are carried out in the first embodiment on detection of different events belonging to a predetermined set of events. This set includes here three events, namely, the (periodic) feedback by the servers of cluster 2 of measures of use of their resources, the positioning of tasks and the reservation of resources arising therefrom on servers, and the release of resources. on the servers. The values updated during the P2 process are preferably taken into account as soon as they are available by the PI process for the processing of requests received after each update.

We will now describe each of these processes in more detail.

With reference to FIG. 3A, it is assumed that the device 3 for managing the cluster 2 receives, via its communication means 9 and its reception module 3A, a request REQ for executing a computer task T of the client device 4 (step E10). This REQ request comprises, as mentioned previously, a quantity R_R_T of resource R initially required for the computer task T to be able to execute properly, this quantity R_R_T having been predetermined by the user of the client device 4.

On receipt of the REQ request (T, R_R_T), the device 3 for managing the cluster 2, by means of its determination module 3B, here determines among the servers S1, ..., SK of the cluster 2 which are those which are able to execute the task T, taking into account the needs R_R_T in resource R specified by the user. In other words, the determination module 3B determines which server (s) has a sufficient amount of resource R to execute the computer task T. It is noted that this determination need not be exhaustive, and a a reduced list of servers to a single server may be sufficient.

To this end, for each server Sk of cluster 2, k = l, ..., K (initialization step E20, test step E50 then incrementation step E60), the determination module 3B determines whether the server Sk checks the following inequality (eql) (test step E30):

R_R_T + EU_R (Sk) <CL_R (Sli) (eql) in which:

- CL_R (Sk) designates the maximum capacity of the Sk server for the IT resource R (i.e. maximum quantity of R resource available on the Sk server). This maximum capacity is raised here periodically to the management device 3 by each server Sk; and - EU_R (Sk) designates an estimate of a quantity already used of the IT resource R on the server Sk.

If the server Sk checks the inequality (eql) (answer yes to the test step E30), it is identified by the determination module 3B as a server capable of executing the task T (step E40).

Otherwise (answer no to test step E30 and test step E50), the determination module 3B tests a new server from among the servers not yet tested in cluster 2 (answer no to test step E50 and incrementation step E60).

In the first embodiment described here, once all the servers Sk, k = l, .., K of cluster 2 have been tested (answer yes in test step E50), the selection module 3C selects a server SkO among the servers identified as being able to execute the computer task T taking into account its needs R_R_T in resource R initially supplied by the user (step E70). No limitation is attached to the criteria used by the cluster management device 3 to select the SkO server. This choice can be made for example in relation to the other resources required for the execution of the task and available on the SkO server, in relation to geographic constraints, etc.

Following this selection, the device 3 for managing the cluster 2 orders (or triggers) the launch of the task T on the selected SkO server (step E80). In other words, it positions the task T on the SkO server and reserves on this server the quantity R_R_T of resource R for execution.

The device 3 for managing the cluster 2 thus proceeds for each IT task submitted to it.

We note that the asynchronous process P1 which has just been described with reference to FIG. 3A and which is implemented by the device 3 for managing the cluster 2 is similar to the treatments implemented by the cluster management systems of the state of the art, and in particular by the Borg and Hadoop Yarn systems mentioned above. In the treatments implemented by these prior art systems, the estimate EU_R (Sk) of the quantity of resource R already used on the server Sk is taken equal, in equation (eql), to the quantity Q_R (Sk) of resource R initially reserved (ie allocated) by the management system on the server Sk for the execution of the tasks previously received and still being processed by the server Sk, that is to say always positioned on the Sk server (this number of positioned tasks is assumed to be equal to J0, J0 designating an integer greater than 0). This quantity Q_R (Sk) is based on the declarations of the needs R_R_Tj in resource R supplied by the users for these previous tasks Tj, j = l, ..., JO, in other words, it is equal to the sum of the needs R_R_Tj, j = l, ..., JO in resource R declared by the users at the origin of these J0 tasks Tj, j = l, ..., JO, again:

Q_R (Sk) = Σ '°, R.RTj.

In other words again, the state of the art management systems such as those mentioned above apply, for the allocation of a resource R to an incoming computer task T which is submitted to them, the following inequality (eql '):

R.R.T + Q.R (Sk) <CL.R (Sk) (eq 1 ')

According to the invention, the device 3 for managing the cluster 2 differs from the state of the art management systems in that certain values used in the equation (eql), and in particular the value EU_R (Sk) and / or the limit capacity value CL_R (Sk), are updated on detection of three predetermined events (defining a predetermined set of events within the meaning of the invention) (step F10). These three events include:

- obtaining measurements of actual (current) use of the IT resource R on the servers Sk, k = l, ..., K of the cluster 2 server;

The reservation by the cluster management device 3 for an IT task of a quantity of the IT resource R on a server Sk, k = l, ..., K of the cluster 2; and - the release by a computer task of a quantity of the computer resource R on a server Sk, k = l, ..., K of the cluster 2.

This update is illustrated in FIG. 3B now described (continuous process P2) for the quantity EU_R (Sk) (variable V within the meaning of the invention).

Taking into account the three aforementioned events allows the device 3 for managing the cluster 2 to have more precise information on the quantity of resource R available on each server during the allocation of the computer tasks via the PI process. Each event designated by EV detected among these three events results in an update of the current value of the variable V considered (here EU_R (Sk)). We note in the following description V (n) the current updated value of the variable V, n designating an integer greater than or equal to 0 indexing the update (in other words, each detected event increments the integer n (initialization steps F00 then incrementation F20)). Thus, when this variable is EU_R (Sk), the current value of this variable is noted EU_R (n, Sk). It is the same for the other variables considered in the description. It is noted that a so-called current value designates an available value or which one modifies at a time considered said current (that is to say at a time in progress). In the case of a measurement, it may have been carried out at a previous instant (in particular in the case of periodicity of the measurement); if this is the case, the current measurement designates the measurement that is available at the current time when it is used.

According to the invention, according to the detected event, an update different from the current value of the variable V is envisaged. We will now detail these three updates.

As indicated above, in the first embodiment described here, it is assumed that the feedback of the measurements made on the servers Sk, k = l, ..., K to the device 3 for managing the cluster 2 is carried out periodically, according to a Tupd period. For the sake of simplification, the same period Tupd is considered here for all the servers of the cluster, however the invention also applies to distinct periods from one server to another. In addition, the invention also applies when the measurements are raised at predetermined times, not necessarily spaced regularly, or in a completely asynchronous manner.

This periodic obtaining of measurements of actual use of the resource R by the management device 3 is detected by the cluster management device 3 (answer yes in step F10 with FV = UPD-U). Such a measurement can be carried out by each Sk server by means of tools known per se, such as for example by means of a “top” command launched by the operating system of each Sk server periodically. We denote by U_R (n, Sk) the measurement of the real use of the IT resource R carried out on the server Sk and available at the current time at the level of the device 3 for managing the cluster 2.

Each of the measurements U_R (n, Sk), k = l, ..., K obtained by the device 3 for managing the cluster 2 is used by the device 3 for managing the cluster as a measure of the current actual use of the IT resource R on the Sk server.

From this measurement, the device 3 for managing the cluster 2, via its updating module 3E, updates, in the first embodiment described here, the current value of the estimate EU_R (Sk) of the quantity already used of the computer resource R on the server Sk intended to be taken into account by the management device 3 during step E30 of the process PI (step F30).

In the first embodiment described here, the update module 3E calculates the current value EU_R (n, Sk) according to the following equation (eq2) (step F30):

EU_R (n, Sk) = (1 - a) EU_R (n - l, S / c) + ai / _R (n, Sk) (eq2) where a denotes a real weighting factor between 0 and 1. We note that in this equation, EU_R (0, Sk) is initialized to the quantity of the resource R reserved on the server Sk at an instant t = 0 for all the tasks positioned on this server by the device 3 for managing the cluster 2 .

We note that equation (eq2) makes it possible to update the estimate of the quantity already used of the IT resource R on the server Sk by calculating the value EU_R (n, Sk) at the current time from a value vpast (n) = EU_R (n- l, Sk) calculated (update) at a previous instant (in this case here, during its previous update). It includes an exponential moving average parameterized by the factor a. The closer the factor a is chosen to 0, the longer the cluster management device 3 takes into account the quantity of resources reserved and delays the moment when it uses free resources from the servers which seem "unduly" reserved. On the contrary, the closer the factor a is chosen to 1, the more quickly the cluster management device 3 forgets these resources reserved by the tasks, taking the risk that these resources are in fact necessary for the tasks for which they had been reserved but used later. The choice of factor a therefore results from a compromise between these different considerations.

Equation (eq2) advantageously makes it possible to adjust (ie to update) the level EU_R (n, Sk) of use of the resource R used by the device 3 for managing cluster 2 for the allocation of resources to IT tasks entrusted to it, taking into account the actual use of resources on the servers of cluster 2. This allows, if a task executing on a Sk server of cluster 2 requires less R resource to execute than the needs in this resource initially announced by the user in the request to execute this task, to allocate to other tasks the resource R not used on the Sk server, or conversely to stop the allocation of resource R to incoming tasks if the Sk servers have no more R resources available. A better allocation is therefore obtained by the device 3 for managing the resources of the cluster 2, while ensuring proper execution of the tasks on the cluster.

In the first embodiment described here, the updating of the current value of the variable EU_R (Sk) by means of the equation (eq2) is carried out periodically, with each feedback of measurements received by the device 3 of cluster management 2. In other words, the event of obtaining measurements (EV = UPD-U) by the management device 3 occurs periodically.

In addition to this periodic update, in accordance with the invention, other events (EV = RESA and EV = FREE in FIG. 3B) are likely to trigger asynchronous updates of the current value of the variable V = EU_R (Sk) for the different Sk servers, k = l,.,., K. More specifically, these asynchronous updates are triggered by the update module 3E each time that it detects one and / or the other of the following two events for one of the Sk servers:

The reservation by the device 3 for managing a quantity of the resource R on the server Sk, for the execution of a computer task Tnew (following the application of the process PI shown in FIG. 3A by the device 3 management on the Tnew task) (EV branch = RESA in FIG. 3B). The quantity of resource reserved for this task Tnew by the management device 3 on the server Sk corresponds to the needs R_R_Tnew in resource R initially required by the user in the execution request for this task transmitted to the management device 3. In the following description, this event is designated by EV = RESA;

- The release by a computer task Told of a quantity of the resource R on the server Sk, due in particular to the completion of this task Told or to its branch interruption EV = FREE in FIG. 3B). In the following description, this event is designated EV = FREE.

For each Sk server concerned for which an EV = RESA event is detected by the management device 3 (via its 3D detection module) in association with a Tnew task (answer yes to the detection step F10, EV branch = RESA) , the current value of the variable EU_R (Sk) is updated by the update module 3E according to the following equation (eq3-RESA) (step F40):

EU_R (n, Sk) = EU_R (n - 1, S k) + R_R_Tnew (eq3-RESA)

The device 3 for managing the cluster 2 being at the origin of the positioning (allocation) of the tasks on the various Sk servers of the cluster, it can immediately inform via its selection module 3C, the 3D detection module of the reservation of new resources on a Sk server.

For each Sk server concerned for which an EV = FREE event is detected by the management device 3 in association with a Told task (answer yes to the detection step F10, branch EV = FREE), the current value of the variable EU_R (Sk) is updated by the update module 3E according to the following equation (eq3-FREE) (step F50):

EU_R (n, Sk) = EU_R (n - 1.5 / c) - (1 - a ') ^nold R_R_Told (eq3-FREE) where nold indexes the number of updates to EU_R performed on detection of the event UPD -U (new measurement back to device 3 for managing the real use of the resource) from the moment when the quantity of resource R_R_Told was reserved for the Told task on the Sk server until its release, the quantity R_R_Told corresponding to the initial resource requirements R declared by the user in the request to execute the Told task addressed to the management device 3.

In equation (eq3) -FREE, the factor (1 - a) ^nold R_R_Told takes account of the balance of resource R that is released by the Told task at the server Sk due to the end of its execution by the Sk server. Note that taking this balance into account is of special interest for tasks whose execution is short-lived (for example less than the Tupd update period).

The asynchronous updates of the variable EU_R (Sk) thus carried out upon each detection of an event EV = RESA and / or EV = FREE aim to take account of the tasks allocated to each Sk server by the device 3 for managing the cluster 2 and tasks that ended on each Sk server between two measurements feedback from the Sk server (and therefore between two periodic updates of the EU_R (Sk) variable).

It is noted that, as mentioned previously, the EV = RESA events can be easily detected by the device 3 for managing the cluster 2, the latter being in charge of allocating resources to the computer tasks entrusted to it. EV = FREE events are detected here by the Sk servers on which the Told tasks had been distributed (positioned) and relayed by these servers to the cluster management device 3 (accompanied by the corresponding nold indices). The reception by the device 3 of management of a notification from a Sk server of the completion of a Told task constitutes a detection of an EV = FREE event for this Sk server within the meaning of the invention. 3 The management device then has the information necessary to evaluate the factor (l - a ') ^nold R_R_Told.

The current value EU_R (n, Sk) updated from the variable EU_R (Sk) is used as soon as it is available (ie as soon as it is updated) by the determination module 3B of the device 3 for managing the cluster 2 as part of the PI process, and more specifically in inequality (eql) to select a server in cluster 2 able to execute a new incoming task received as of this update.

The steps F10 to F50 of the process P2 are then reiterated each time an event EV = UPD-U, RESA or FREE is detected by the 3D detection module.

In a first variant of this first embodiment, the updating module 3E uses during step F30 to calculate the current value EU_R (n, Sk) the following equation (eq2 ') in place of the equation (eq2):

EU_R (n, Sk) = max ((l - a) EU_R (n - 1.57c) + aU_R (n, Sk), U_R (n, Sk /) (eq2 ') where max (x, y) denotes the function which provides the maximum value among the x and y values The update processes described in steps F40 and F50 also remain identical to that described previously.

This first alternative embodiment makes it possible to take into account more quickly the real use of resources on the Sk server.

In a second variant of the first embodiment illustrated in FIG. 3C, the updating module 3E uses to calculate the current value EU_R (n, Sk) the following equation (eq2) in place of the equation (eq2 ) (step G30):

EU_R (n, Sk) = DR_R (n, Sk) + U_R (n, Sk) (eq2) where DR_R (n, Sk) denotes a smoothed current value of the quantity of the IT resource reserved for at least one IT task per the management device 3 on the server Sk at the current time. The smoothed value DR_R (n, Sk) is calculated here by the update module 3E during step G30, in the second variant, according to the following equation (eq4):

DR_R (n, Sk) = (1 - (i) DR_R (n - i.Sk) (eq4) where β is a real weighting factor between 0 and 1. We note that the choice of weighting factor β corresponds to a compromise similar to that mentioned above for the choice of the weighting factor a (ie speed of forgetting the reserved resources versus actual use observed of these resources). In this equation, DR_R (0, Sk) is initialized to the quantity of the resource R reserved on the server Sk at time t = 0 for all the tasks positioned on this server.

It is assumed here that the value of the variable DR_R is updated periodically according to the same period Tupd as the feedback of the measurements to the management device 3.

However this assumption is not limiting in itself, and the update of the value of DR_R according to the equation (eq4) can be carried out according to another period, asynchronous compared to the rise of U_R (n, Sk). Each update of the value of DR_R (Sk) then results in an update of the value of EU-R (Sk). It is therefore another event (event EV = UPD-D shown in dotted lines in FIG. 3C), in the same way as the events EV = UPD-U, EV = RESA and EV = FREE described above, likely in the second variant described here to be detected by the 3D detection module and to trigger an update of the current value of the variable V (= EU-R (Sk) here) (step G40).

In addition, in this second variant, when an EV = RESA or EV = FREE event is detected, the value of DR_R is updated according to the equations (eq5-RESA) for the detection of an EV = RESA event (step G50) and (eq5-FREE) for the detection of an EV = FREE event (step G60) as follows:

D_R (n, Sk) = D_R (n - ï, Sk) + R_R_Tnew (eq5-RESA) and

D_R (n, Sk) = D_R (n - l.Sk) - (1 - a) ^nol “R_R_Told (eq5-FREE) where nold indexes the number of updates of D_R carried out on detection of the UPD-U event (new measurement returned to the device 3 for managing the real use of the resource) from the moment when the quantity of resource R_R_Told has been reserved for the task Told on the server Sk until its release, the quantity R_R_Told corresponding to initial requirements for resource R declared by the user in the request to execute the Told task addressed to the management device 3.

The current value of EU-R is then updated during steps G50 and G60 according to equation (eq2) where U_R (n, Sk) designates the current value available at the level of the device 3 for managing the usage measurement resource R.

In a third variant of the first embodiment, each measurement of the actual use of the computer resource R is carried out on each server Sk of the cluster 2, k = l, ..., K, for each separate task Tj executed by this Sk server using the IT resource R, j = O, ..., Jl, J1 designating an integer greater than 0 enumerating the tasks positioned on the Sk server during the measurement. We are limited here only to the tasks Tj executed by the server Sk and managed by the device 3 for managing the cluster 2 (that is to say which were launched on the server Sk on instruction from the device 3 for managing the cluster 2 after execution of the PI process).

U_R (n, Sk, j) denotes the measurement of the actual use of the computer resource R carried out on the server Sk for the task Tj at the end of the time interval t (n-l). Each measurement U_R (n, Sk, j) carried out on a server Sk is sent at the end of the time interval t (n) to the device 3 for managing the cluster 2, and obtained during step F20 by the 3D module for obtaining the device 3 for managing the cluster 2. Each measurement U_R (n, Sk, j) is used by the device 3 for managing the cluster as a measure of the actual use of the computer resource R on the server Sk by the task Tj for the time interval t (n).

On the basis of these measurements, the device 3 for managing the cluster 2, via its updating module 3E, evaluates in the third variant, during step F30, an estimate EU_R (n, Sk, j) of the quantity already used of the computer resource R on each server Sk by each computer task Tj managed by the device 3 for managing the cluster 2 and executed on the server Sk using the following equation (eq6):

EU_R (n, Sk, j) = (1 - a) EU_R (n - l, Sk, j) + aU_R (n, Sk, j) (eq 6) or the following equation (eq6 '):

EU_R (n, Sk, j) = max ((l - a) EU_R (n - l.Sk.f) + aU_R (n, Sk, j), U_R (n, Sk, j)) (eq 6 ') in which a is an actual weighting factor between 0 and 1, chosen as described previously in the other alternative embodiments. We note that in this third variant, EU_R (nOj, j) is initialized to the quantity R_R_Tj of IT resource R initially required for the execution of the task Tj, nOj designating the update index corresponding to the allocation of the computer task Tj to the server Sk by the management device 3.

Then, in the third variant described here, during step F30, the updating module 3E evaluates from the values EU_R (n, Sk, j) obtained for each of the tasks Tj the quantity EU_R (n, Sk ) according to the following equation (eq2 '):

Jlact

EU_R (n, Sk) = EU_R (n, Sk, f) (eq2'j 7 = o or according to the following equation (eq2 ⁽⁴⁾ ):

EU_R (n, Sk) = max ^ ^a EU_R (n, Sk, j), Σ ^ U_R (n, Sk, j)) (eq2 ⁽⁴⁾ ) in which Jlact designates the number of tasks positioned on the Sk server ( that is to say allocated to the server Sk by the management device 3) among the J1 tasks for which measurements have been reported to the management device 3.

In the first embodiment and in its three variants described above, it is the device 3 for managing the cluster 2 which, from the measurements reported by the servers Sk, k = l, ..., K of the cluster 2, updates the value of the quantity EU_R (Sk) used by the module 3B for determining the device 3 for managing the cluster 2 to select a server of the cluster to execute an incoming task.

In a second embodiment, each server Sk of cluster 2 is a server according to the invention. He understands :

- a measurement module, configured to measure, periodically here, the actual use of its IT resource R;

- a detection module configured to detect a predefined set of events (UPD-U, RESA and FREE); and an update module configured to update at least one current value of a variable denoted V chosen from an estimate of an amount already used of the computer resource on the server and a maximum capacity of the server for the resource. IT, said update module being activated on each detection by the event detection module from among the predefined set of events and configured for:

o if the detected event is a measurement (U_R (n)) of a current actual use of the computer resource on the server, update the current value of the variable V using said measurement of the current actual use of the computer resource on the server;

o if the detected event is a reservation by the management device for a computer task of a quantity of the computer resource on the server, update the current value of the variable V using said quantity of the reserved computer resource; and if the detected event is a release by a computer task of a quantity of the computer resource on the server, updating the current value of the variable V by using said quantity of the released computer resource; and a transmission module, configured to transmit to the management device said current value of the updated variable V.

More precisely, in the second embodiment, on each detection of an EV event (EV = UPD-U, RESA, FREE or UPD-D as in the first embodiment) each Sk server estimates via its update module update the current quantity EU_R (n, Sk) already used of its IT resource R. It can do this for example in a similar or identical way to what was described previously in the first embodiment and implemented by the management device 3 with reference to FIGS. 3B and 3C.

Then, from the quantity EU_R (n, Sk) thus estimated by the server Sk, the update module of the server Sk updates the value of its maximum capacity CL_R (Sk) for the resource R and for the time interval t (n) according to the following equation (eq7):

CL_R (n, Sk ~) = CL_R (Sk) + Q_R (n, Sk) - EU (n, Sk) (eq7) where Q_R (n, Sk) designates the quantity of resource R initially reserved by the management device 3 on the Sk server for the execution of tasks positioned on the Sk server at the current update time. This quantity Q_R (n, Sk) is given by, taking into account the needs R_R_Tj, in resource R declared by the users at the origin of these tasks:

Q_R (n, Sk) = Σ ^ Γ ^ -β-Ό · where J2 (n) designates the number of tasks Tl, ..., TJ2 (n) positioned on the Sk server at the current update time .

The current value of the maximum capacity CL_R (n, Sk) of the server Sk for the resource R is then raised (periodically for example here) by the server Sk to the device 3 for managing the cluster 2, for example using the communication protocol already used in IT systems based on a Hadoop Yarn management system.

The device 3 for managing the cluster 2 then uses the updated value CL_R (n, Sk) of the maximum capacity CL_R (Sk) of the server Sk in the equation (eql ') mentioned above to determine, for an incoming task T submitted to it by the servers Sk, k = l, ..., K of cluster 2 capable of performing this task.

It should be noted that this second embodiment does not require any modification of the processing carried out today by the state of the art management systems (i.e. they use the equation (eql ')). Thus, in this second embodiment, the device 3 for managing the cluster 2 can be a Hadoop Yarn management system or a Borg management system already known. Only the Sk servers of cluster 2 must be modified to implement the invention. These modifications can affect all or part of the servers in cluster 2.

In the two embodiments described here, as well as in their variants, an exponential moving average has been considered in order to update the quantity EU_R (n, Sk). However, this assumption is not limiting in itself, and other types of smoothing or sliding averages can be considered to calculate the quantity EU_R (n, Sk). Thus, for example, in an alternative embodiment, we can consider an arithmetic average of the actual use measurements reported on a sliding window comprising N time intervals preceding the time interval t (n).

In yet another variant, it is possible to consider a weighted moving average MMP (n, Sfc) of the actual use measurements reported on a sliding window comprising N time intervals preceding the current update time, such as for example;

MMDC CkA - ^N ' ^u - ^R (n.SIC) + (N'-1) U_R (n — l, Sk) + · + U_R (nN ^, + l, 5fc)

MMP {n, Sk) - w '+ (n'-i) + - + i where N' is the maximum of the number of measurements U_R (n, Sk) available preceding the measurement U_R (n, Sk) and N. This average MMP (n, Sk) is written in an equivalent manner, after having obtained N measurements U_R (m, Sk), in the form:

M MP (n, S k) = MMP (η - l.Sk) + ^u - ^R ^ ^Sk ^^ ⁿ We note that in this weighted average, the measurements U_R (n, Sk) are all the more significant as 'They are recent.

The value EU_R (n, Sk) can then be calculated according to the following equation, on receipt of a new measurement U_R (n, Sk) (detection of an EV event = UPD-U):

n-l

EU_R (n, Sk) = MMP (n, Sk) + Σ Σ (k + N - n) * R_R_Tj k = n — N + l ^a i ~ ^k

N 'dj> n where aj indexes the start of the Tj task and dj the end of the Tj task on the Sk server.

The EU_R (n, Sk ') value can also be updated as follows when an EV = RESA event is detected:

FÎ / J? (N, Sk) = EU_R (n, Sk) + R_R_Tnew

These examples are only given by way of illustration and still other examples can be envisaged to update the value EU_R (n, Sk).

Claims (17)

1. Method for updating, for at least one server (Sk) of a cluster of servers, with at least one current value of a variable denoted V chosen from an estimate of an already used quantity of a resource computing on the server or a maximum capacity of the server for said computing resource, said at least one current value being intended to be taken into account by a server cluster management device for selecting, on receipt of a request for execution of 'an IT task associated with a predetermined quantity of said IT resource (R) initially required for its execution, a so-called target server of the cluster to execute this IT task, said method being such that said update is carried out on each detection (F10 ) an event (EV) from a predefined set of events and: - if the detected event (UPD-U) is a obtaining of a measurement (U_R (n)) of a current actual use of the computer resource on the server, said update (F30, G30) of the value current of the variable V uses said measure of the current actual use of the computer resource on the server; - if the detected event (RESA) is a reservation by the management device for an IT task of a quantity of the IT resource on the server, said update (F40, G50) of the current value of the variable V uses said quantity of the reserved computer resource; and - if the detected event (FREE) is a release by a computer task of a quantity of the computer resource on the server, said update (F50, G60) of the current value of the variable V uses said quantity of the IT resource released. 2. The updating method according to claim 1, in which if the detected event (UPD-U) is the obtaining of the measurement of the current actual use of the computer resource on the server, said updating of the current value of the variable V also uses at least one value of the variable V denoted Vpast updated at a previous instant. 3. The updating method according to claim 2, in which if the detected event (UPD-U) is the obtaining of the measurement of the current actual use of the computer resource on the server, during said updating. the measurement of the current actual use of the computer resource is weighted by a real factor a between 0 and 1, and said at least one value Vpast is weighted by a factor equal to (1-a). 4. The updating method according to claim 3, in which the variable V update is the estimate EU_R of the quantity already used of the computer resource on the server, and if the event detected is the obtaining of the measurement. of the current actual use of the computer resource on the server, said current value denoted EU_R (n) of the variable V is updated according to the equation: EU_R (n) = max ((l - a) EU_R (n - 1) + aU_R (n), U_R (nj) where n denotes an integer greater than 1, EU_R (n - 1) is the value of the estimate of the quantity already used of the computer resource on the server updated at a previous instant, and U_R (n) denotes the measurement obtained from the current actual use of the computer resource on the server. 5. The updating method according to claim 1, in which, if the event detected is the obtaining of the measurement of the current actual use of the computer resource on the server, this measurement corresponds to a use of the computer resource. on the server by one or more IT tasks running on the server and all managed by the cluster management device. 6. Method for updating according to claim 5, in which the variable V updated is the estimate EU_R of the quantity already used of the computer resource on the server, and if the event detected is the obtaining of the measurement. the actual current use of the IT resource on the server: The measurement of the current actual use of the computer resource on the server comprises a plurality of values denoted U_R (nJ), j = l, ... J, associated respectively with a plurality of computer tasks Tj, j = l, ..J, positioned by the cluster management device on the server, J and n designating integers greater than or equal to 1, each value U_R (nJ) corresponding to a measure of the actual use of the IT resource on the server by the task Tj; - said update (F30) includes: a first evaluation step for each task Tj, of a current value EU_R (n, j) of an estimate of an amount already used of the computer resource on the server by the computer task Tj according to the equation: EU_R (n, f) = max ((l - a) EU_R (n - IJ) + aU_R (nJ), U_R (njj) or according to the equation: EU_R (nJ) = (1 - a) EU_R (n - l, y) + aU_R (nJ) where: EU_R (n - l, y) designates a value evaluated at a previous instant of the estimation of the quantity already used of the IT resource on the server by this IT task Tj; and a is a real factor between 0 and 1; ο a second step of evaluation of the current value noted EU_R (ri) of the estimation of the quantity already used of the computer resource on the server according to the equation: EU) _R (n) = ^ EU_R (n, j) / = i or according to the equation: EU_R (n) = maxl '^ EU_R (n, j) U_R (n, f) V = i 7 = 1 / where J denotes the number of computer tasks positioned on the server by the cluster management device at the time of the second evaluation step. 7. The updating method according to claim 1, in which the updated variable V is the estimate EU_R of the quantity already used of the computer resource on the server, and if the event detected is the obtaining of the measurement. of the current actual use of the computer resource on the server, said current value denoted EU_R (n) of the variable V is updated according to the equation: EU_R (n) = U_R (n) + DR_R (n) where U_R (n) denotes a current measure of the actual use of the computer resource on the server, and DR_R (n) denotes a current value of a quantity of the IT resource reserved for at least one IT task by the management device on the server. 8. Updating method according to claim 7, further comprising: A step for updating the current value DR_R (n) of the quantity of the IT resource reserved for at least one IT task by the management device on the server is obtained according to the equation: DR_R (n) = (1 - p) DR_R (n - 1) where β denotes a real factor between 0 and 1 and DR_R (n - 1) denotes a quantity of the IT resource reserved by the management device on the server obtained at a previous instant; - a detection step (G10) of the update (UPD-D) of the current value DR_R (n); and - an update step (G40) of the current value of the variable EU_R according to the equation: EU_R (n) = U_R (n) + DR_R (n). 9. Updating method according to claim 7 or 8 further comprising: - if the detected event is a reservation by the management device for a computer task of a quantity R_R_Tnew of the computer resource on the server, a step of updating the current value DR_R (n) of the quantity of the IT resource reserved for at least one IT task by the management device on the server using said quantity R_R_Tnew and a quantity DR_R (n - l) of the IT resource reserved by the management device on the server obtained at a previous instant, - if the detected event is a release by an IT task of an amount R_R_Told of the IT resource on the server, a step of updating the current value DR_R (n) of the amount of the IT resource reserved for at least a computer task by the management device on the server using said quantity R_R_Told and quantity DR_R (n - 1); and in which said updating of the current value of the variable EU_R uses said quantity of the computer resource reserved according to the equation: EU_R (n) = U_R (n) + DR_R (n). 10. Method for managing computer tasks comprising, on reception (E10) by a device (3) for managing a cluster (2) of servers (S1, S2, ..., SK) of a request (REQ ( T, R_R_T)) of execution of a computer task (T), this computer task being associated with a predetermined quantity (R_R_T) of a computer resource (R) initially required for its execution: A step of determining (E40), by the cluster management device, of at least one server of the cluster capable of executing the computer task, this step of determining using the predetermined quantity of the computer resource initially required for the execution from the IT task, a maximum capacity of the server for the IT resource and an estimate of an already used amount of the IT resource on the server; A step of selection (E70), by the cluster management device, of a server to execute the IT task among said at least one server of the determined cluster; said method being characterized in that, for at least one server of the cluster, at least one current value of a variable denoted V chosen from the estimate of the quantity already used of the computer resource on the server and the maximum capacity of the server for the computer resource is updated according to an updating method according to any one of claims 1 to 9. 11. The management method as claimed in claim 10, in which the updating method is implemented by the device (3) for managing the cluster. 12. Management method according to claim 10 wherein the updating method is implemented by said at least one server (Sk), said management method further comprising a step of transmitting said current value of the variable V said at least one server updates the cluster management device. 13. Computer program (PROG) comprising instructions for the execution of the steps of the update method according to any one of claims 1 to 9 or instructions for the execution of the steps of the management method according to claim 10 when said program is executed by a computer. 14. recording medium readable by a computer on which a computer program is recorded comprising instructions for the execution of the steps of the updating method according to any one of claims 1 to 9 or instructions for the execution of the steps of the management method according to claim 10. 15. Management device (3) for a duster (2) of servers (S1, ..., SK) comprising: - a reception module (3A); A determination module (3B), activated on reception by the reception module of a request for execution of an IT task, said IT task being associated with a predetermined quantity of an IT resource initially required for its execution, the determination module being configured to determine at least one server of the duster capable of performing the computer task using the predetermined quantity of the computer resource initially required for the execution of the computer task, a maximum capacity of the server for the computer resource and an estimate of a quantity already used of the computer resource on the server; - a selection module (3C) configured to select a server to execute the IT task among said at least one server in the cluster determined by the determination module; said management device being characterized in that it further comprises: - a detection module configured to detect a predefined set of events; and - an update module activated for at least one server of the cluster, and configured to update at least one current value of a variable denoted V chosen from the estimate of the quantity already used of the computer resource on the server and the maximum capacity of the server for the IT resource, said update module being activated on each detection of an event from the predefined set of events by the detection module, and configured for: o if the detected event is a measurement (U_R (n)) of a current actual use of the computer resource on the server, update the current value of the variable V using said measurement of the current actual use of the computer resource on the server; o if the detected event is a reservation by the management device for a computer task of a quantity of the computer resource on the server, update the current value of the variable V using said quantity of the reserved computer resource; and o if the detected event is a release by a computer task of a quantity of the computer resource on the server, update the current value of the variable V using said quantity of the computer resource released. 16. Server (Sk) belonging to a cluster (2) of servers managed by a management device (3), said server being able to execute at least one IT task managed by said management device, said at least one IT task using during its execution at least one computer resource of the server, said server comprising: - a measurement module configured to measure an actual use of the IT resource on the server; - a detection module configured to detect a predefined set of events; and an update module configured to update at least one current value of a variable denoted V chosen from an estimate of an amount already used of the computer resource on the server and a maximum capacity of the server for the resource. IT, said update module being activated on each detection by the event detection module from among the predefined set of events and configured for: o if the detected event is a measurement (U_R (n)) of a current actual use of the computer resource on the server, update the current value of the variable V using said measurement of the current actual use of the computer resource on the server; o if the detected event is a reservation by the management device for a computer task of a quantity of the computer resource on the server, update the current value of the variable V using said quantity of the reserved computer resource; and o if the detected event is a release by a computer task of a quantity of the computer resource on the server, update the current value of the variable V using said quantity of the computer resource released; and a transmission module, configured to transmit to the management device said current value of the updated variable V. 17. Computer system (1) comprising: - a cluster (2) of servers comprising a plurality of servers, each server comprising at least one IT resource allowing the execution of IT tasks; - a device (3) for managing the server cluster; wherein said cluster management device is in accordance with claim 15 and / or at least one server in the cluster is in accordance with claim 16. 1/3 v