KR102035394B1 - Automatic methodology for optimizing the techniques of process live migration - Google Patents

Abstract

The present invention relates to a method of migrating a process of a node to another node capable of performing the same function when a node controlling a specific function fails. Specifically, how to prioritize various functions according to predefined criteria, create a routing table using priorities, and migrate the process of a failed node to other nodes based on the routing table and the proximity between nodes. Related.

Description

AUTOMATIC METHODOLOGY FOR OPTIMIZING THE TECHNIQUES OF PROCESS LIVE MIGRATION}

The present invention relates to a method of migrating a process in charge of a specific system to another system capable of performing the same function in the environment in which systems having various functions are connected, when a specific system fails.

In the present specification, the system generally refers to computing equipment including desktop and lab tam, embedded controllers, cellular equipment, and the like. Autonomous restoration systems and technologies are emerging to autonomously restore a system so that if a problem occurs in the system, the running process is not affected.

Dual systems are widely used to automate process restoration. Dual system refers to the operation of two systems to prevent process interruption due to system failure. It is a method used to obtain high stability of the system, and is used for air control, military, broadcasting equipment maintenance, and banking management. Dual systems require at least twice the physical hardware and equipment as any hardware or equipment fails, requiring another hardware and equipment to handle the same process in order to prevent disconnection. Therefore, if the remaining hardware and equipment also fails, the process may be interrupted, which is a one-time solution.

One of the systems that can be used for autonomous restoration is a cloud system that uses virtual machines as opposed to physical hardware and devices as a replacement device for failed hardware. A cloud system refers to a system that uses a virtual machine created through at least a portion of a remote server through a communication network. As the process progresses, information about the system can be backed up to the cloud in real time and used to prevent process disruption. However, when using a cloud system, since data exists in a third space, if the security of the network is compromised, the data can be easily lost and there is an additional cost in that there is only a place for backup. In addition, a temporary process of migration requires a permanent space in the cloud, resulting in waste of resources. Container technology is convenient because it uses a cloud system, but has similar disadvantages.

Process migration is another method used for autonomous process restoration. Process migration is a technique used for load balancing purposes to prevent overloading a particular system. It is gaining attention in autonomous recovery systems in that the process itself can be moved from one system to another.

One embodiment according to the present invention may provide a process migration method for automatically migrating a process by automatically selecting optimized backup systems without manually designating a backup system.

The process migration method according to the present invention comprises the steps of prioritizing each of a plurality of functions; Determining a failed first node among a plurality of nodes controlling each of the plurality of functions; Determining at least one node of the plurality of nodes capable of performing the function of the first node instead; Determining a second node from which to migrate a process of the first node from among the at least one node based at least in part on the priority or the proximity between the first node and the at least one node; And migrating a process of the first node to the second node.

Prioritizing each of the plurality of functions may include prioritizing each of the plurality of functions based at least in part on big data.

Prioritizing each of the plurality of functions may include prioritizing each of the plurality of functions based at least in part on artificial intelligence.

Prioritizing each of the plurality of functions may include prioritizing each of the plurality of functions based on power consumption of each of the plurality of functions.

The determining of the second node may include performing a process of the first node among the at least one function using at least partially a routing table generated based on the plurality of nodes, the plurality of functions, and the priority. Determining a second node to migrate.

The process migration method according to the present invention can migrate the process by selecting optimized backup systems without manually designating the backup system.

1 is a diagram illustrating a system for performing process migration, according to an exemplary embodiment.
2 is a diagram for explaining a routing table.
3 is a flowchart illustrating a process migration method.

Hereinafter, specific embodiments of the various embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, these specific embodiments do not limit or limit the present invention. Regardless of the reference numerals in the drawings, the same reference numerals denote the same components, and redundant descriptions are omitted.

1 is a diagram illustrating a system for performing process migration, according to an exemplary embodiment.

Referring to FIG. 1, a system for performing process migration includes a central node 110, a network 120, CPS nodes 130 and 140, and functions 150 and 160 necessary for the system. The central node 110 performs overall control of the system.

The central node 110 may give priority to the functions 150, 160. Priority may be automatically assigned to each function through a user's predetermined importance determination criteria, may be manually assigned by the user, and each of automatic grant and manual grant may be performed at least partially. For example, when a criterion is specified such that the priority of a function having a large power consumption is low, the central node 110 may grasp the power consumption of each function and give priority to each function. For example, when the importance of each function is designated as a priority standard in a vehicle accident, the central node 110 may give the vehicle engine cooling function a higher priority than the air conditioner function.

The central node 110 may give priority to the functions 150 and 160 based at least in part on the big data. For example, the central node 110 collects big data about the cause of the car accident, big data about the price of auto parts, and judges the importance of each function of the car based on the big data, Each function can be given priority. For example, the central node 110 may give the vehicle engine cooling function a higher priority than the air conditioner function based on the big data on the cause of the vehicle accident.

The central node 110 may prioritize the functions 150, 160 based at least in part on artificial intelligence. For example, the central node 110 may include, for example, supervised learning, unsupervised learning, semi-supervised learning, through an artificial neural network (ANN). Artificial intelligence, which has performed at least one of reinforcement learning, may be used to calculate and prioritize the probability of thinking in the absence of each function. For example, the central node 110 may calculate that the probability of a vehicle accident is higher when the vehicle engine cooling function is absent than when the vehicle engine cooling function is not provided, and may give the vehicle engine cooling function a higher priority than the air conditioner function.

The central node 110 may monitor the CPS nodes 130 and 140 to determine whether the functions 150 and 160 are being performed. If a specific function is not performed due to a failure, the central node 110 may migrate a process of a CPS node that was controlling a specific function not performed due to the failure to another CPS node. For example, when the vehicle engine cooling function 150 is not performed due to a failure, the central node 110 performs a process of the first CPS node 130 that controls the vehicle engine cooling function and controls the air conditioner function. Migrate to CPS node 140. The second CPS node 140 may perform only the vehicle engine cooling function without controlling the air conditioner function, but may control not only the vehicle engine cooling function but also the air conditioner function through the function distribution optimized by the central node 110. .

In one example, if there are several other CPS nodes that can replace the CPS node that was controlling a specific function that is not performed due to a failure, the priority can be migrated. For example, when the vehicle engine cooling function fails, the central node 110 may migrate the process in consideration of the priority of the air conditioner function and the external air injection function that may replace the vehicle engine cooling function. When the priority is given based on the probability of the accident, if the central node 110 determines that the air conditioner function has a higher priority than the external air injection function, the CPS node performing the external air injection function is the process of the vehicle engine cooling function. You can take over.

The central node 110 may control process migration based on the proximity between the CPS nodes controlling the functions 150 and 160. For example, when a failure of the vehicle engine cooling function occurs, the central node 110 performs a vehicle engine cooling function with each of the second CPS node 140 controlling the air conditioner function or the third CPS node performing the external air injection function. The proximity of the first CPS node 130 that has been controlled may be determined, and the migration may be performed based on this. Proximity can be calculated by measuring packet forwarding speed and physical distance between CPS nodes. For example, if the packet forwarding speed between the first CPS node 130 and the second CPS node 140 is faster than the packet forwarding speed between the first CPS node 130 and the third CPS node, the first CPS node 130 The vehicle engine cooling function that was controlled by the second CPS node 140 may be controlled. That is, the process of the first CPS node 130 may be migrated to the second CPS node 140.

The central node 110 may migrate with partial consideration of priority and proximity. For example, you can select a CPS node to control the failure function and migrate the process through formulas that both priority and adjacency are considered variables and can be identified numerically.

The central node 110 may re-prioritize each function at any time or at a defined time. For example, even if the central node 110 gives the vehicle engine cooling function the highest priority when driving in a rain-free environment, the central node 110 reprioritizes the priority at any time or at a defined time. This gives the wiper function the highest priority when driving in a rainy environment.

The central node 110 and the CPS nodes 130, 140 communicate over the network 120. The central node 110 or the CPS nodes 130 and 140 may communicate with an external server, terminal, sensor, or the like through the network 120. The CPS nodes 130 and 140 may control the physical device through a network or a communication path capable of transmitting and receiving electrical signals.

The network 120 may include, for example, at least one of a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), and a global area network (GAN). It may include one.

The virtual physics system (CPS) is a system in which physical devices are physically converted into virtual spaces in order to facilitate control of physical devices and to minimize time and space wasted in controlling the physical devices. The CPS node according to the present invention is a control unit which is embedded in a server, a personal computer, a portable communication terminal, and the like to virtualize and store information on a physical device and control the physical device based on the information.

The CPS nodes 130, 140 may control the physical devices such that the respective functions 150, 160 are performed. For example, as shown in FIG. 1, the first CPS node 130 may control the cooling device such that, for example, the vehicle engine cooling function is performed, and the second CPS node 140 may be, for example, an air conditioner. The air conditioner may be controlled to perform the function. The CPS node may receive signals from the central node 110 to control specific physical devices to perform other functions. For example, as shown in FIG. 1, the second CPS node 140 may control the air conditioner to perform a vehicle engine cooling function by receiving a signal from the central node 110. The CPS node can control a specific physical device to perform several functions in addition to one function. For example, the second CPS node 140 may not only control the air conditioner to perform the vehicle engine cooling function but also control the air conditioner to perform the air conditioner function using at least some of the available tasks.

The central node 110 may control the migration based on the routing table. The routing table provides a guide for migrating processes for specific functions in the event of a system failure. The routing table is generated based on the priority of each function assigned by the central node 110. The central node 110 uses the routing table to migrate the process of the CPS node that was controlling the failed function to the CPS node controlling the replaceable function. For example, the central node 110 uses the priority-based routing table to control the process of the first CPS node 130 that was controlling the failed vehicle engine cooling function 150 to control the air conditioner function. Migrate to node 140.

2 is a diagram for explaining a routing table.

Referring to FIG. 2, there are functions 211 through 216 on the vertical axis of the routing table and CPS nodes 221 through 226 on the horizontal axis. In the grayed diagonal direction in the routing table, each of the CPS nodes represents functions under control before a particular function fails. The first CPS node controls the first function, the second CPS node controls the second function, the third CPS node controls the third function, the fourth CPS node controls the fourth function, and the fifth The CPS node controls the fifth function, and the sixth CPS node controls the sixth function. The number in parentheses next to each function indicates the priority of the function, and the number in parentheses next to each CPS node indicates the priority of the function controlled by that node. Priority is given only to the function, but for convenience of explanation, priority is indicated next to the node controlling the function.

The central node 110 assigns the first rank to the first function, the second rank to the second function, the third rank to the third function, the fourth rank to the fourth function, the fifth rank to the fifth function, and the sixth rank to the sixth function. . The central node 110 migrates a process to a CPS node that controls the replaceable function if a particular CPS node is unable to perform a function due to a failure. For example, when a failure occurs in the first CPS node 221, since the first function 211 is not performed, the central node 110 selects the fifth CPS node 225 that controls the replaceable function. The priority of the function controlled by the first CPS node 221 and the function controlled by the fifth CPS node 225 may be compared. The fifth function 215 controlled by the fifth CPS node 225 has a lower priority than the first function 211 controlled by the first CPS node 221, so that the fifth function 215 is controlled by the fifth CPS node 225. The process of the CPS node 221 may be migrated.

If you have more than one CPS node that controls the replaceable feature, you can migrate based on the priority of the replaceable CPS nodes. For example, if the second CPS node 222 fails, the central node 110 migrates the process because the second function 212 is not performed. In this case, both the third CPS node 223 and the fourth CPS node 224 may replace the second function 212. The central node 110 determines that the third function 213 controlled by the third CPS node 223 has a higher priority than the fourth function 214 controlled by the fourth CPS node 224, and the second function The process of the CPS node 222 may be migrated to the fourth CPS node 224 having a lower priority.

If the third CPS node 223 fails, since the third function 213 is not performed, the central node 110 migrates the process. In this case, both the first CPS node 221 and the sixth CPS node 226 may replace the third function 213. The central node 110 determines that the first function 211 controlled by the first CPS node 221 has a higher priority than the sixth function 216 controlled by the sixth CPS node 226, and the third Processes of the CPS node 223 may be migrated to the sixth lower priority CPS node 226.

If the fourth CPS node 224 fails, the central node 110 migrates the process because the fourth function 214 is not performed. In this case, the first CPS node 221 may replace the fourth function 214. The first function 211 controlled by the first CPS node 221 has a higher priority than the fourth function 214 controlled by the fourth CPS node 224, but may perform the fourth function 214. Since the CPS node is only the first CPS node 221, in this case, the CPS node may be migrated to a high priority CPS node.

If the fifth CPS node 225 fails, the central node 110 migrates the process because the fifth function 215 is not performed. In this case, both the third CPS node 223 and the sixth CPS node 226 may replace the fifth function 215. The central node 110 determines that the third CPS node 223 has a higher priority than the sixth CPS node 226, and the process of the fifth CPS node 225 has a sixth CPS node 226 having a lower priority. ) Can be migrated.

If the sixth CPS node 226 fails, the central node 110 migrates the process because the sixth function 216 is not performed. In this case, both the second CPS node 222 and the fourth CPS node 224 may replace the sixth function 216. The central node 110 determines that the second CPS node 222 has a higher priority than the fourth CPS node 224, and determines that the process of the sixth CPS node 226 has a lower priority than the fourth CPS node 224. ) Can be migrated.

The central node 110 may migrate processes based on the proximity between the CPS nodes in addition to the routing table. For example, as shown in FIG. 2, when the second CPS node 222 fails, the central node 110 does not immediately migrate the process to the lower priority and fourth CPS node 224. The migration is performed in consideration of the proximity between the second CPS node 222 and the third CPS node 223 or the fourth CPS node 224. For example, if the packet transmission rate between the second CPS node 222 and the third CPS node 223 is faster than the packet transmission rate between the second CPS node 222 and the fourth CPS node 224, the second CPS node ( Since the proximity between the 222 and the third CPS node 223 is high, the central node 110 may migrate the process to the third CPS node 223 in consideration of this.

3 is a flowchart illustrating a process migration method.

Referring to FIG. 3, the process migration method according to an embodiment of the present invention includes assigning a priority to each of the plurality of functions (S310). For example, in an automobile system, priority may be given to an automobile engine cooling function, an air conditioning function, a transmission function, a brake function, and the like.

Functions may be prioritized based at least in part on big data. For example, it collects big data about the cause of car accidents, big data about auto parts price, and judges the importance of each function of the car based on big data and prioritizes each function according to the importance. Can be given. For example, the vehicle engine cooling function may be given a higher priority than the air conditioner function based on the big data on the cause of the vehicle accident.

Functions may be prioritized based at least in part on artificial intelligence. For example, through an artificial neural network (ANN), for example, supervised learning, unsupervised learning, semi-supervised learning, reinforcement learning Using artificial intelligence that has performed at least one of the above, priorities may be calculated by calculating an accident probability when each function is absent. For example, the vehicle engine cooling function may be given a higher priority than the air conditioner function by calculating that the vehicle accident probability is higher than that without the air conditioner function.

Priority may be given to the functions automatically according to the criteria defined above, but the priority of each function may be determined according to the user's discretion.

After operation S310, monitoring a plurality of nodes that control each of the plurality of functions may include determining a first node in which a function is not performed due to a failure (S320). For example, when the vehicle engine cooling function is not performed, the vehicle engine cooling node controlling the function is determined.

After operation S320, the method may include determining at least one node that may control a function controlled by the first node (S330). For example, the air conditioner node or the fan node may be determined as a node that can be controlled in place of the vehicle engine cooling function.

After operation S330, the method may include determining a second node to which the process of the first node is to be migrated based on the priority or the proximity between the first node and the at least one node (S340).

When there are several other nodes that can replace the first node that performed a specific function not performed due to a failure, the second node may be determined in consideration of the priority. For example, when a failure occurs in the vehicle engine cooling function, the second node may be determined in consideration of the priority of the air conditioner function and the external air injection function that can replace the vehicle engine cooling function. When priority is given based on the probability of an accident, if the air conditioner function is given a higher priority than the external air injection function, the node performing the lower priority external air injection function performs the vehicle engine cooling function. Can be determined to migrate the process.

The second node may be determined based on the proximity between the first node and the at least one node. For example, when a failure of the vehicle engine cooling function occurs, the proximity between each node controlling the air conditioner function or a node controlling the external air injection function and a node controlling the vehicle engine cooling function is determined, and a second node is determined. Can be. Proximity can be calculated by measuring packet forwarding speed and physical distance between nodes. For example, if the packet forwarding speed between the node controlling the vehicle cooling function and the node controlling the external air injection function is faster than the packet forwarding speed between the node controlling the vehicle cooling function and the node controlling the air conditioner, the vehicle engine cooling function may be The node controlling the external air injection function can control it. That is, in order to migrate the process of the node that controlled the vehicle cooling function, the node controlling the external air injection function may be determined as the second node.

Not only can the second node be determined based on each of the priority or the neighbor, but the second node can be determined based in part on both the priority and the neighbor. For example, the second node may be determined through a formula that can consider both the priority and the adjacency and can be confirmed numerically.

After operation S340, the method may include migrating a process of the first node to the second node (S350). For example, when a failure of the vehicle engine cooling function, which is a function of the first node, the process of the first node may be migrated to the second node so that the node controlling the air conditioner function, which is the second node, controls the vehicle engine cooling function. .

Although the present invention has been described in detail with reference to the limited embodiments as described above, the present invention is not limited to the above-described embodiments. Those skilled in the art to which the present invention pertains can see and change the claims and the description of the invention, and can be easily changed, modified and carried out to the implementation within the scope of the claims of the present invention.

Claims (5)

Prioritizing each of the plurality of functions;
Determining a failed first node among a plurality of nodes controlling each of the plurality of functions;
Determining at least one node of the plurality of nodes capable of performing the function of the first node instead;
Determining a second node from among the at least one node to migrate a process of the first node based on the priority and the adjacency between the first node and the at least one node; And
Migrating a process of the first node to the second node;
Including,
Prioritizing each of the plurality of functions
Priority is given to each of the plurality of functions based on the importance of each of the plurality of functions, and re-prioritization of each of the plurality of functions at any time or at a defined time based on the surrounding environment. and,
The adjacency
Calculated by measuring a packet forwarding speed and a physical distance between the first node and the at least one node,
Determining the second node is
Determine a second node to migrate a process of the first node from among the at least one function by using a routing table generated based on the plurality of nodes, the plurality of functions, and the priority;
Determining the second node is
In the case where there are a plurality of the at least one node, a node having a lower priority among the priorities of functions controlled by the node is determined as the second node, or a node having a faster packet forwarding speed or a physical distance closer to the second node. How to migrate processes that determine by node. The method of claim 1,
Prioritizing each of the plurality of functions
Prioritizing each of the plurality of functions based at least in part on big data
Process migration method comprising the. The method of claim 1,
Prioritizing each of the plurality of functions
Prioritizing each of the plurality of functions based at least in part on artificial intelligence
Process migration method comprising the. The method of claim 1,
Prioritizing each of the plurality of functions
Prioritizing each of the plurality of functions based on power consumption of each of the plurality of functions
Process migration method comprising the. delete

Images