EP3732573A1 - Procede de gestion de panne dans un reseau de noeuds et partie de reseau de noeuds associee - Google Patents
Procede de gestion de panne dans un reseau de noeuds et partie de reseau de noeuds associeeInfo
- Publication number
- EP3732573A1 EP3732573A1 EP18842811.4A EP18842811A EP3732573A1 EP 3732573 A1 EP3732573 A1 EP 3732573A1 EP 18842811 A EP18842811 A EP 18842811A EP 3732573 A1 EP3732573 A1 EP 3732573A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- node
- nodes
- storage medium
- network
- state
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/16—Error detection or correction of the data by redundancy in hardware
- G06F11/20—Error detection or correction of the data by redundancy in hardware using active fault-masking, e.g. by switching out faulty elements or by switching in spare elements
- G06F11/202—Error detection or correction of the data by redundancy in hardware using active fault-masking, e.g. by switching out faulty elements or by switching in spare elements where processing functionality is redundant
- G06F11/2023—Failover techniques
- G06F11/2025—Failover techniques using centralised failover control functionality
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/14—Error detection or correction of the data by redundancy in operations
- G06F11/1402—Saving, restoring, recovering or retrying
- G06F11/1415—Saving, restoring, recovering or retrying at system level
- G06F11/1438—Restarting or rejuvenating
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/14—Error detection or correction of the data by redundancy in operations
- G06F11/1446—Point-in-time backing up or restoration of persistent data
- G06F11/1458—Management of the backup or restore process
- G06F11/1464—Management of the backup or restore process for networked environments
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/16—Error detection or correction of the data by redundancy in hardware
- G06F11/20—Error detection or correction of the data by redundancy in hardware using active fault-masking, e.g. by switching out faulty elements or by switching in spare elements
- G06F11/202—Error detection or correction of the data by redundancy in hardware using active fault-masking, e.g. by switching out faulty elements or by switching in spare elements where processing functionality is redundant
- G06F11/2023—Failover techniques
- G06F11/2033—Failover techniques switching over of hardware resources
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F13/00—Interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
- G06F13/38—Information transfer, e.g. on bus
- G06F13/40—Bus structure
- G06F13/4004—Coupling between buses
- G06F13/4022—Coupling between buses using switching circuits, e.g. switching matrix, connection or expansion network
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/16—Error detection or correction of the data by redundancy in hardware
- G06F11/20—Error detection or correction of the data by redundancy in hardware using active fault-masking, e.g. by switching out faulty elements or by switching in spare elements
- G06F11/202—Error detection or correction of the data by redundancy in hardware using active fault-masking, e.g. by switching out faulty elements or by switching in spare elements where processing functionality is redundant
- G06F11/2038—Error detection or correction of the data by redundancy in hardware using active fault-masking, e.g. by switching out faulty elements or by switching in spare elements where processing functionality is redundant with a single idle spare processing component
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2213/00—Indexing scheme relating to interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
- G06F2213/0026—PCI express
Definitions
- the invention relates to the field of fault management methods in a node network and the domain of node network portions associated with this fault management.
- backups are performed at one or more levels, they are multilevel backups.
- the calculation can be recovered at least partially without having to be completely restarted, precisely thanks to the backups made.
- one or the other level of backup is used to recover the calculation in part or in the majority or almost completely.
- Distributed applications can last much longer than the average network failure time, also called MTBF of a cluster (for "Mean Time Between Failures" in English), so they are likely to be interrupted. In general, they do not have an internal fault-management solution, which can then lead to the loss of local backup data in the event of a physical failure of the compute node, followed by the loss of the entire data set. calculation progress caused by the loss of a single compute node but whose local backup data can no longer be retrieved.
- the different levels of backup range from the most local backup to the most global backup, ranging from the simplest and fastest backup to the most complex, slowest, and most expensive backup, ranging from the most fragile and weakest backup to the most robust and resilient backup.
- the first level Ll which makes a local backup, simple and inexpensive, performed very often, resulting in a minimal loss of computation time during a failure that can be recovered at this first level first Ll,
- the second level L2 which performs a first intermediate backup by duplication on a partner node, less simple and more expensive, resulting in a greater loss of computation time during a failure that can be recovered only at this second level L2 ,
- the third level L3 which makes a second intermediate backup by Reed-Solomon type encoding, even less simple and even more expensive, resulting in a loss even more important in computing time during a failure that can not be recovered that at this third level L3,
- the fourth level L4 which performs a global backup at a file system level, complex and very expensive, resulting in a really significant loss of computing time during a failure that can be recovered only at this fourth level L4.
- the backup is more and more robust and resilient, but it also becomes more and more complex and expensive. That's why the backup of the first level It is performed often, the backup of the second level L2 less often, the backup of the third level L3 even less often, the backup of the fourth level L4 relatively rarely. Therefore, statistically, when a failure occurs, the last coherent state that can be retrieved is very recent at the first level L1, less recent at the second level L2, even less recent at the third level L3, older at the fourth level L4. Consequently, the volume of work lost is very low at the first level L1, relatively limited at the second level L2, notable at the third level L3, and more important at the fourth level L4.
- the object of the present invention is to provide a fault management method in a node network at least partially overcoming the aforementioned drawbacks.
- the invention aims to provide a fault management method in a network of nodes improving the compromise between efficiency on the one hand and cost and complexity on the other hand, for at least one level of backup considered.
- the aim of the invention is to provide a fault management method in a network of nodes that has an efficiency similar or comparable to that of an intermediate level backup, preferably that of the first intermediate backup, advantageously that of the second level. L2, for a cost and complexity similar or comparable to those of a local level backup, preferably those of the first level Ll.
- the invention proposes to make the link between a storage medium and its redirection node to another node, so as to have the backup performed on the storage medium when the node is failing, possibly through a copy at a level Neighbor node, but without making backup copy at a neighboring node for most nodes that are not failing or for all nodes that are not failing.
- the link between a storage medium and its node is not realized directly, but indirectly through a network element able to reconfigure this link to link this storage medium to another node when the previous node becomes faulty.
- This network element links several nodes to their respective storage media, each node being connected to its associated storage medium (or possibly to its associated storage media).
- a backup is available despite the failure of this node, resulting in a similar efficiency to the second level L2, but no operation is more complex than a simple local backup for the second node.
- most or all nodes that are not failing which is the majority or even the vast majority of the nodes of the network performing the calculation considered, resulting in a cost and complexity at least comparable if not similar to those of the first level Ll .
- the cost of the second level L2 is saved, while the ability to restart an application from the backups of the first level L1 is maintained in the event of failure of a computing node.
- the copy to a neighboring node traditionally performed at the second level L2 is here not performed during the execution of the application in a preventive manner for all the compute nodes, but only in case of failure and only for the failed nodes after occurrence of the breakdown. The copy is then made only to restart the application with the missing data, uploads from the local backup.
- this backup of cost and complexity similar or comparable to a local backup has the effectiveness of an intermediate backup, which greatly improves the value for money of this backup.
- this backup of cost and complexity similar to or comparable to a local backup having the efficiency of an intermediate backup preferably makes it possible to replace both a conventional local backup and one or more conventional intermediate backups, by not keeping more than the global backup last resort, to handle the most severe failures, most of the failures can now be managed by the local backup - intermediate proposed by the invention, the latter having the effectiveness of a backup intermediate practically at the cost of a local backup.
- the present invention proposes a fault management method in a network of nodes, comprising, for each considered node of all or part of the nodes of the network performing the same calculation: first, a local backup step of the network; state of this node considered, at the level of a storage medium of this node considered, the link between this storage medium and this node can be redirected from this storage medium to another node, then a recovery step : the node considered if it is not faulty, from the local backup of the state of this node, or an operational node different from the considered node, if the considered node is failing, from recovering the local backup of the state of this node, by redirecting said link between the node in question and its storage medium so as to connect said storage medium to said node.
- the local backups of these nodes considered, used for the recovery steps are consistent with each other so as to correspond to the same state of this calculation.
- the present invention also proposes a part of a network of nodes, comprising: a switch, a plurality of nodes of a group of nodes performing the same calculation in this network of nodes, several storage media respectively attached to these nodes via the switch, these storage media being structured and arranged to locally store the state of the nodes to which they are respectively attached, this switch being structured and arranged for change at least one attachment between a failed node and its storage medium which will then become attached to another node of said plurality of nodes.
- this fault management method allows the optimization and the resilience of the data backup of an application likely to encounter a failure and the restart of this application using the data saved. .
- this fault management method offers the possibility of migrating a storage medium from the failed computing node by means of the switch, thus making it possible to directly exploit the local data saved from a node of calculating down in a time that depends only little on the amount of data saved.
- this fault management method provides fault tolerance that extends in absolute terms to a large number of compute nodes as long as it exists, on each computing node that falls in failure, a calculation node attached to the same switch that can resume the calculation of the failed computing node. It is then possible to restart from the last local backup, even if a large number of the compute nodes fail, as long as each failed compute node has a switch and a neighboring non-faulty computing node connected at the same switch.
- the invention comprises one or more of the following features which can be used separately or in partial combination with one another or in total combination with one another, applied to one or the other of the objects of the invention. invention mentioned above.
- each considered non-faulty node there is no step of recovery by another node, the local backup of the state of this node considered not faulty.
- the operational node different from the defective node concerned is a spare node.
- non-failing nodes which already have their own computation task to perform, are not loaded in addition to performing this operation of recovering a faulty neighbor node. A possible overload of these non-faulty nodes is then avoided.
- said redirection of said link between the node in question and its storage medium so as to connect said storage medium to said operational node is achieved by a change of routing in a switch connecting several nodes to their storage media.
- this redirection is performed by a simple operation performed at a network element that is reliable and controlled.
- all the nodes restarting steps are synchronized with each other, so as to restart all said nodes in the same calculation state.
- the recovery step comprising a substep of migration, to the operational node or to the spare node, storage medium of the local backup of the state of the failed node, which relates , this storage medium of the local backup of the state of the failed node, at a predetermined location of the file tree of this operational node or of this spare node, this operational node or this spare node then read, at this predetermined location, the link to the local backup of the state of the failed node, during the recovery step of this operational node or of this spare node, no copy, of the local backup of the state of the node failing, being realized, at another node.
- the spare node and the failed node that it replaces both belong to the same compute blade.
- the recovery step comprises: firstly, a substep of migration, to an intermediate node, of the storage medium of the local backup of the state of the failing node, which links, this storage medium of the local backup of the state of the failed node, at a predetermined location of the file tree of this intermediate node, then, a substep of copy, the local backup of the state of the node failing, the storage medium attached to the intermediate node to the storage medium of the operational node or the spare node, the operational node or this spare node then read its storage medium, during its recovery stage.
- the copy sub-step is not necessary and can be avoided by moving the storage medium to the right place in the spare node's file tree as in the first embodiment described above, or by simply changing the configuration of the the fault tolerance library.
- This feedback is instantaneous and allows to avoid an additional copy that could take a little time depending on the amount of data to save.
- the replacement node and the failed node that it replaces belong to different calculation blades.
- the redirection of the link between the support and the node can be performed even in the event of a major failure and even relatively generalized at the level of a whole computing blade.
- the migration sub-step changes the attachment of the storage medium of the local backup of the state of the failed node by passing through a switch to which the failed node was attached and its storage medium. local backup of the state of the failed node, but without going through the failed node itself.
- the redirection can be performed even in case of complete physical failure of the failed node.
- the change of attachment is achieved by sending a command to the switch, this command passing through one of the nodes attached to the switch by a management port.
- this switch is a PCIe switch
- this switch is particularly advantageous because it is particularly suitable for communicating devices with each other without having to go through a microprocessor, thus being able to bypass a faulty computing node for example.
- the use of this PCIe switch allows the storage medium, for example a storage disk, to be reattached, containing the local backups of the failed computing node to a spare computing node. This operation is fast and does not require a systematic copy of local backup data, and in particular not for non-faulty compute nodes.
- 3 to 10 nodes are attached to the same switch.
- the switch can easily manage this small group of nodes for which only one spare node seems to have to be sufficient.
- the fault management method also includes, for all or part of the nodes of the network performing the same calculation: a step of global backup of all of these nodes, performed less often than all the local backup steps of these nodes.
- the local-intermediate backup proposed by the invention for managing the vast majority of faults in a simple and efficient manner, and on the other hand the more complex and expensive overall backup but reserved for a minority of severe failures, an excellent compromise between overall complexity and overall efficiency of the fault management method proposed by the invention is achieved.
- the fault management method proposed by the invention then allows restarting the application, following a physical failure on a node, even complete, and this in most cases, starting from all the local backups, more recent and less expensive, instead of having to restart remote backups that are very often significantly older, some rarer cases of failure sometimes requiring the use of remote global backup.
- the ability to recover local data from the failed compute node allows you to restart the application from most recent local backups in most cases.
- the network comprises between 1 and 5 spare nodes for 100 nodes performing the same calculation.
- this fault management method does not include any other type of local backup of the state of these nodes.
- the storage media are flash memories.
- the memories used are simple, fast and permanent.
- these flash memories are memories NVMe ("Non Volatile Memory Express" in English).
- the memories used are particularly well suited to communicate with a PCIe switch ("Peripheral Component Interconnect Express" in English).
- this switch being structured and arranged to change at least one attachment between a faulty node and its storage medium, on command external to the switch.
- said plurality of nodes comprises between 3 and 10 nodes.
- the switch can easily manage this small group of nodes for which only one spare node seems to have to be sufficient.
- this switch is a PCIe switch.
- this switch is particularly advantageous because it is particularly adapted to be able to communicate with each other peripherals without having to go through a microprocessor.
- the storage media are flash memories, preferably NVMe memories.
- the memories used are particularly well suited to communicate with a PCIe switch.
- the network of computation nodes comprises at least 1000 computation nodes, advantageously at least 5000 computation nodes, still more advantageously at least 10000 computation nodes, making the fault management method according to the invention all the more interesting. Because the complete loss of a current calculation then becomes all the more critical as the network is large.
- fault tolerance is all the more important that an application runs on a cluster that consists of a larger number of compute nodes. The greater the number of processors, memory and other devices, the greater the probability that a failure will occur before the end of the execution.
- Applications that are intended to run on this type of platform will use fault tolerance libraries that allow them to save (“checkpoint” in English) the data needed for a restart (“restart” in English). in a state as close as possible to the state that existed just before the failure. Thus, these applications are not forced to restart the calculation from the beginning.
- FIG. 1 diagrammatically represents an example of a network part including a group of nodes and their storage media connected to each other by a PCIe switch, according to one embodiment of the invention, at which part of the network, an example of a fault management method according to the invention can take place.
- FIG. 1 diagrammatically represents an example of a network part including a group of nodes and their storage media connected to each other by a PCIe switch, according to one embodiment of the invention, at which part of the network, an example of a fault management method according to the invention can take place.
- This part of the network comprises several calculation nodes 2, three calculation nodes 21, 22 and 23, in the example of FIG. 1, as well as several storage supports 3, three storage media 31, 32 and 33, example of Figure 1.
- connection 7 is a 4-way connection.
- Connection 8 is a 4-way connection.
- connection 9 is a 2-way connection, here not yet used, each connection 9 being on one side attached to one of the ports 13, 16 or 19 and remaining free on the other side.
- Each PCIe connection 7 respectively connects one of the computing nodes 21 to 23 to one of the ports 0, 2 or 8 (numbering the electronic switch PLX 8733, but another PCIe switch can be used), respectively referenced 11, 14 or 17 in Figure 1, of the switch 1.
- Each PCIe connection 8 respectively connects one of the storage media 31 to 33 to one of the ports 1, 3 or 9 , respectively referenced 12, 15 or 18 in Figure 1, of the switch 1.
- the connection 7, the ports 0 and 1, respectively referenced 11 and 12 in Figure 1, of the switch 1, the connection 8 together form a link 4 connecting the computation node 23 at its support 33.
- the connection 7, the ports 2 and 3, respectively referenced 14 and 15 in FIG. 1, of the switch 1, the connection 8 together form a link 5 connecting the computation node 22 to its support 32.
- the connection 7, the ports 8 and 9, respectively referenced 17 and 18 in Figure 1, the switch 1, the connection 8 together form a link 6 connecting the compute node 21 to its support 31.
- the PCIe connections 7 , 8 or 9, can be
- the computation node 23 is attached to the default management or management port, that is to say that it is through it that the outgoing sendings of the group of calculation nodes 2 and the receptions from from the outside of the computing node group 2. In the event of failure of this computing node 23, it is replaced by the computing node 22 which is attached to the redundant management or management port, which computing node 22 becomes attached to the new port of management or effective management.
- the attachment of the storage medium 31 is reconfigured, that is to say that the storage medium 31 will cease to be connected to its calculation node 21 of attachment by default, but will become connected to the calculation node 23, which is attached to the default management port, will be able to trace the local backup of the calculation state of the failed computing node 21, since the storage medium 31, to another spare computing node outside the group of calculation nodes 21 to 23, this other computing node then taking over the calculation task interrupted at the failed computing node 21, from the local backup of the computing state of the computation node 21 sent up from the storage medium 31.
- the port 9 (numbering of the electronic switch PLX8733, but another PCIe switch can be used) referenced 18 (in FIG. 1), instead of remaining permanently connected to the port 8 referenced 17 as before the failure of the computing node 21, will be, at least temporarily connected to the port 0 referenced 11, to allow the computing node 23 to read, in the storage medium 31, the saved data representative of the calculation state of the node of calculation 21, just before or shortly before, its failure.
- the local backup in the storage medium 31, of the calculation state of the calculation node 21 before its failure, will be able to go back to the computing node 23 and thus be exploited to restart the calculation with a very good compromise between simplicity of the backup and efficiency of the recovery.
- the computing node 23 is itself a spare node, it can take over the computation task interrupted by the failed computing node 21.
- the calculation node 22 If it is the node 23 that becomes faulty, it is first replaced by the calculation node 22 as attached to the management port, and the calculation node 22, as attached to the new management port, performs the operations previously performed by the computing node 23 if it had not become defective.
- the management of the switch 1 is now described by a manual scenario clarifying and explaining the various operations to be performed, such as for example the migration of the storage medium 31 from a failing calculation node 21 to a spare calculation node 23, the transfer data, restarting the application.
- Application-side data management is abstracted by the FTI library.
- the executed application is provided in the examples of the FTI library: hdf.exe.
- This application is launched on two compute nodes. It will perform local backups on NVMe storage disks attached to these compute nodes by the PCIe switch at regular intervals as well as a global backup on a NFS server ("Network File System" in English) less frequently .
- NFS server Network File System
- the migration of the NVMe storage disk from the failed node to another node will then allow the transfer of data from the last local backup of that failed node to the spare node. Once this transfer is made, the application can be restarted and resumes the calculation at the last local backup of the two compute nodes instead of the last older global backup of the network part.
- a failure recovery is performed with the local backup data of an application MPI ("Message Passing Interface" in English) of a computing node crashing.
- the data is retrieved by migrating the storage medium of the failing compute node to a compute node adjacent to the same compute blade. This data is then transmitted to a second spare node which will resume the calculation.
- MPI Message Passing Interface
- the advantage of this failover is that it allows the application to restart from local backups of all compute nodes. These cheaper local backups are also newer most of the time, and are at worst only as recent as global backups. In addition, the amount of data transferred to resume the calculation will be much lower than in the case of a restart from a global backup.
- the configuration of the FTI library has been modified ["config.fti" in the “examples” directory], to allow execution on two different compute nodes:
- the local backup directory ["/ localckpt /"] is an assembly of the SSD storage disk.
- Solid State Disk available on each compute node, the storage disk S0 (referenced 31 in FIG. 1) on the calculation node N1 and the storage disk S1 (referenced 32 in FIG. 1) on the compute node N2:
- the computing node N 1 is considered to be down, the storage disk S0 attached to it will be migrated to another computing node, here the computing node N3 (referenced 23 in Figure 1).
- the compute node N3 is the node that has the switch administration port.
- the first step is to send the command that allows to reroute the reset PCI ("reset" in English), the last argument being the number of the faulty or faulty node:
- This command can be executed from neighboring nodes or the management or management node.
- the recovered saved data is then transmitted to the spare computing node in charge of replacing the failed computing node N1. It can be the same node as the node N3 in charge of the recovery of the data saved on the storage disk S0.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Quality & Reliability (AREA)
- Mathematical Physics (AREA)
- Computer Hardware Design (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Hardware Redundancy (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| FR1763237A FR3076004B1 (fr) | 2017-12-27 | 2017-12-27 | Procede de gestion de panne dans un reseau de noeuds et partie de reseau de noeuds associee |
| PCT/FR2018/053535 WO2019129986A1 (fr) | 2017-12-27 | 2018-12-21 | Procede de gestion de panne dans un reseau de noeuds et partie de reseau de noeuds associee |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP3732573A1 true EP3732573A1 (fr) | 2020-11-04 |
Family
ID=62143294
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP18842811.4A Pending EP3732573A1 (fr) | 2017-12-27 | 2018-12-21 | Procede de gestion de panne dans un reseau de noeuds et partie de reseau de noeuds associee |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US11249868B2 (fr) |
| EP (1) | EP3732573A1 (fr) |
| FR (1) | FR3076004B1 (fr) |
| WO (1) | WO2019129986A1 (fr) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110324375B (zh) * | 2018-03-29 | 2020-12-04 | 华为技术有限公司 | 一种信息备份方法及相关设备 |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100199128A1 (en) * | 2009-01-30 | 2010-08-05 | International Business Machines Corporation | Fanout connectivity structure for use in facilitating processing within a parallel computing environment |
| US20120185727A1 (en) * | 2011-01-17 | 2012-07-19 | Doe Hyun Yoon | Computing system reliability |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS6054052A (ja) | 1983-09-02 | 1985-03-28 | Nec Corp | 処理継続方式 |
| US5696895A (en) * | 1995-05-19 | 1997-12-09 | Compaq Computer Corporation | Fault tolerant multiple network servers |
| US7434096B2 (en) * | 2006-08-11 | 2008-10-07 | Chicago Mercantile Exchange | Match server for a financial exchange having fault tolerant operation |
| JP5068056B2 (ja) * | 2006-10-11 | 2012-11-07 | 株式会社日立製作所 | 障害回復方法、計算機システム及び管理サーバ |
| CN102037709B (zh) | 2008-05-21 | 2015-08-12 | 艾利森电话股份有限公司 | 刀片集群交换中心服务器中的资源池化 |
| JP6056509B2 (ja) * | 2013-01-30 | 2017-01-11 | 富士通株式会社 | 情報処理装置および情報処理装置の制御方法 |
| US9026846B2 (en) * | 2013-03-15 | 2015-05-05 | Lsi Corporation | Data recovery in a raid controller by offloading contents of DRAM to a flash module on an SAS switch |
| JP5903524B2 (ja) * | 2013-03-15 | 2016-04-13 | 株式会社日立製作所 | 計算機切替方法、計算機システム、及び管理計算機 |
| US9619389B1 (en) * | 2013-07-11 | 2017-04-11 | Unigen Corporation | System for a backward and forward application environment compatible distributed shared coherent storage |
| US9921926B2 (en) * | 2014-01-16 | 2018-03-20 | Hitachi, Ltd. | Management system of server system including a plurality of servers |
| US9280432B2 (en) * | 2014-03-21 | 2016-03-08 | Netapp, Inc. | Providing data integrity in a non-reliable storage behavior |
| US9588853B2 (en) * | 2014-06-05 | 2017-03-07 | International Business Machines Corporation | Automatic management of server failures |
-
2017
- 2017-12-27 FR FR1763237A patent/FR3076004B1/fr active Active
-
2018
- 2018-12-21 EP EP18842811.4A patent/EP3732573A1/fr active Pending
- 2018-12-21 WO PCT/FR2018/053535 patent/WO2019129986A1/fr not_active Ceased
- 2018-12-21 US US16/958,202 patent/US11249868B2/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100199128A1 (en) * | 2009-01-30 | 2010-08-05 | International Business Machines Corporation | Fanout connectivity structure for use in facilitating processing within a parallel computing environment |
| US20120185727A1 (en) * | 2011-01-17 | 2012-07-19 | Doe Hyun Yoon | Computing system reliability |
Non-Patent Citations (2)
| Title |
|---|
| See also references of WO2019129986A1 * |
| ZIZHONG CHEN ET AL: "Fault tolerant high performance computing by a coding approach", PROCEEDINGS OF THE TENTH ACM SIGPLAN SYMPOSIUM ON PRINCIPLES AND PRACTICE OF PARALLEL PROGRAMMING , PPOPP '05, 1 January 2005 (2005-01-01), New York, New York, USA, pages 213, XP055001673, ISBN: 978-1-59-593080-4, DOI: 10.1145/1065944.1065973 * |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2019129986A1 (fr) | 2019-07-04 |
| US20210073091A1 (en) | 2021-03-11 |
| US11249868B2 (en) | 2022-02-15 |
| FR3076004B1 (fr) | 2021-03-19 |
| FR3076004A1 (fr) | 2019-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9672372B2 (en) | Method for improving mean time to data loss (MTDL) in a fixed content distributed data storage | |
| US10425480B2 (en) | Service plan tiering, protection, and rehydration strategies | |
| EP2092442B1 (fr) | Récupération rapide d'un groupe primaire | |
| US7657780B2 (en) | Enterprise service availability through identity preservation | |
| US20070143374A1 (en) | Enterprise service availability through identity preservation | |
| US10216775B2 (en) | Content selection for storage tiering | |
| US20070168500A1 (en) | Enterprise service availability through identity preservation | |
| CN101501668A (zh) | 通过身份保持的企业服务可利用性 | |
| US20070150526A1 (en) | Enterprise server version migration through identity preservation | |
| US20070143373A1 (en) | Enterprise server version migration through identity preservation | |
| FR2859294A1 (fr) | Systeme de reseau de disques et procede de commande d'informations d'erreur | |
| FR2946769A1 (fr) | Procede et dispositif de reconfiguration d'avionique. | |
| WO2010034920A1 (fr) | Determination et gestion de reseaux virtuels | |
| EP3588294A2 (fr) | Procédé de gestion de panne dans un réseau de noeuds basé sur une stratégie globale | |
| EP3732573A1 (fr) | Procede de gestion de panne dans un reseau de noeuds et partie de reseau de noeuds associee | |
| US20120323869A1 (en) | File State Subset Satellites to Provide Block-Based Version Control | |
| FR3082973A1 (fr) | Procede de gestion de panne dans un reseau de noeuds base sur une strategie locale | |
| EP2799994A1 (fr) | Procédé et dispositif de sauvegarde de données dans une infrastructure informatique offrant des fonctions de reprise d'activité | |
| EP3765959A1 (fr) | Gestion des donnees de configuration pour un serveur multimodule | |
| FR3039347A1 (fr) | Procede de sauvegarde de l'environnement de travail d'une session d'un utilisateur sur un serveur | |
| WO1997030393A1 (fr) | Systeme informatique a transparence de panne pour les applications utilisateur |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
| 17P | Request for examination filed |
Effective date: 20200626 |
|
| AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
| AX | Request for extension of the european patent |
Extension state: BA ME |
|
| DAV | Request for validation of the european patent (deleted) | ||
| DAX | Request for extension of the european patent (deleted) | ||
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: EXAMINATION IS IN PROGRESS |
|
| 17Q | First examination report despatched |
Effective date: 20220413 |
|
| RAP1 | Party data changed (applicant data changed or rights of an application transferred) |
Owner name: COMMISSARIAT A L'ENERGIE ATOMIQUE ET AUX ENERGIES ALTERNATIVES Owner name: BULL SAS |
|
| P01 | Opt-out of the competence of the unified patent court (upc) registered |
Effective date: 20230330 |
|
| RAP3 | Party data changed (applicant data changed or rights of an application transferred) |
Owner name: COMMISSARIAT A L'ENERGIE ATOMIQUE ET AUX ENERGIESALTERNATIVES Owner name: BULL SAS |