Resource management method in data collection system

Abstract

The present disclosure is a data collection system comprising: a plurality of data sources; a server that collects data from the plurality of data sources; and a communication path control device that establishes communication paths between the plurality of data sources and the server. The communication path control device releases a resource of a communication path having a margin in a delay requirement among already established communication paths, and allocates the released resource to a communication path having a high priority.

Description

Resource management methods in data collection systems

[0001] This disclosure relates to communication control for data collection.

[0002] With the spread of IoT (Internet of Things) and remote work, communication traffic continues to increase year after year. When transferring large volumes of data generated from various sensors to a server, conventional communication methods using TCP (Transmission Control Protocol) may experience bottlenecks in the receiving end's communication processing. Furthermore, the increased load on the CPU (central processing unit) may lead to a decrease in receiving processing performance and an increase in power consumption.

[0003] RDMA (Remote Direct Memory Access), which avoids memory copying—a bottleneck in conventional TCP methods—has been proposed. However, RDMA requires a lossless network for data transfer, making it unsuitable for direct use in current wide-area networks. Methods for using RDMA in networks where packet loss occurs include using PFC (Priority Flow Control) or equipping servers with retransmission control mechanisms, but these methods have challenges such as the possibility of deadlocks and increased server load.

[0004] To enable RDMA-based data transfer over wide-area networks, control methods have been proposed that coordinate communication path establishment and RDMA-based data transmission (see, for example, Patent Document 1 and Non-Patent Document 1). For example, Patent Document 1 establishes a communication path that ensures a lossless network with guaranteed bandwidth at the time of data transmission and releases it after transmission is complete. This makes RDMA-based data transfer possible even over wide-area networks.

[0005] In conventional technology, communication paths were established in the order of the communication path allocation requests. As a result, each data source had to wait for its turn to have its communication path established. Consequently, in conventional data acquisition systems, data transmission delays occurred for all data sources waiting for communication path establishment.

[0006] The types of data collected in a data collection system are diverse, and while delays are undesirable for some data, they are acceptable for others. Therefore, in a data collection system that collects data using communication paths, a dedicated resource for high priority is pre-configured to enable rapid path allocation in response to high-priority allocation requests. In this disclosure, the dedicated resource for high priority is referred to as the "dedicated high-priority resource."

[0007] WO2023 / 012878

[0008] Ryohei Tsugami et al., Implementation Evaluation of a Communication Path Utilization Control Method for Wide-Area Multi-Terminal RDMA Data Acquisition, Shinshu University of Science and Technology, B-6-23, May 2022.

[0009] However, even if a high-priority dedicated resource is provided, if that resource is already allocated, even if a new high-priority allocation request with strict delay requirements arrives, the resource cannot be allocated to the high-priority allocation request until the data transmission for other allocation requests, including shared resources available to low-priority allocation requests, has been completed.

[0010] This disclosure aims to enable the allocation of resources to high-priority allocation requests with strict latency requirements, even when both high-priority dedicated resources and shared resources have already been allocated, in a data collection system that collects data using a communication path.

[0011] To achieve the above objective, the data collection system of the present disclosure comprises a plurality of data sources, a server for collecting data from the plurality of data sources, and a communication path control device that, when data is transmitted from one of the plurality of data sources to the server, establishes a communication path between the one data source and the server in accordance with an allocation request from the one data source or the server.

[0012] The data collection system of the present disclosure executes the resource management method of the present disclosure. In the resource management method of the present disclosure, the communication path control device releases resources of an already established communication path that has a grace period for its delay requirement, and allocates the released resources to a high-priority allocation request.

[0013] The communication path control device may comprise a resource management unit that manages high-priority dedicated resources allocated to high-priority allocation requests and shared resources allocable to low-priority allocation requests. In this aspect, the communication path control device may release resources of a communication path that has a grace period for its delay requirement among communication paths using the shared resources.

[0014] When any received high-priority allocation request fails to satisfy the delay requirement, the communication path control device may release resources of an already established communication path that has a grace period for its delay requirement.

[0015] It should be noted that the above respective disclosures may be combined as much as possible.

[0016] According to the present disclosure, in a data collection system that collects data using communication paths, even when both the high-priority dedicated resources and the shared resources have already been allocated, it is possible to allocate resources to a high-priority allocation request with strict delay requirements.

[0017] An example of the system configuration of the present disclosure is shown. An example of the configuration of a communication network is shown. An example of an embodiment of the communication path control device of the present disclosure is shown. An example of a path setting location table is shown. An example of a data transmission sequence is shown. Configuration examples of a request receiving unit and a path setting unit are shown. It is a diagram explaining the problem of the present disclosure. An example of the resource management method of the present disclosure is shown. An example of the operation of the communication path control device is shown. An example of the operation of the communication path control device is shown. An example of the resource management method of the present disclosure is shown. An example of the operation of the communication path control device is shown. An example of the operation of the communication path control device is shown. An example of the operation of the communication path control device is shown.

[0018] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. Note that the present disclosure is not limited to the embodiments shown below. These embodiments are merely illustrative, and the present disclosure can be implemented with various modifications and improvements based on the knowledge of those skilled in the art. In the present specification and the drawings, constituent elements having the same reference numerals are mutually identical.

[0019] (Summary of the present disclosure) FIG. 1 shows an example system configuration of the present disclosure. The data collection system of the present disclosure includes a plurality of data sources 10, a communication path control device 30, and a server 40. The plurality of data sources 10, the communication path control device 30, and the server 40 are connected via a communication network 80.

[0020] The data source 10 is a terminal that holds user-side data such as sensing data acquired by a sensor 50. The server 40 is a terminal that collects data from the data source 10. The communication network 80 is any means that enables packet transfer, and may include a public network regardless of whether it is wired or wireless.

[0021] FIG. 2 shows a configuration example of the communication network 80. The communication network 80 is composed of a plurality of network devices 81. In the present embodiment, the network device 81 used between the data source 10 and the server 40 is predetermined. For example, the data source 10A and the server 40 are connected via network devices 81A and 81B, the data source 10B and the server 40 are connected via network devices 81A and 81C, and the data source 10C and the server 40 are connected via network devices 81D and 81E.

[0022] Figure 3 shows an example configuration of the communication path control device 30. The communication path control device 30 includes a request receiving unit 31, a path setting unit 32, and a path setting location table 33. The request receiving unit 31 receives communication path allocation requests or release requests from each data source 10. The request receiving unit 31 may also receive communication path allocation requests or release requests from the server 40. The path setting unit 32 establishes or releases a communication path in accordance with the communication path allocation request or release request. The path setting location table 33 manages the setting information (establishment / release) of communication paths from each data source 10 to the server 40.

[0023] Figure 4 shows an example of a path setting location table 33. The path setting location table 33 stores information about the network devices 81 that transmit data from each data source 10. For example, in the case of data source 10A, the information about network devices 81 includes identification information for network devices 81A and 81B that connect data source 10A and server 40, and communication path setting information (established / released) for network devices 81A and 81B.

[0024] Figure 5 shows an example of a data transmission sequence. In the data collection system of this disclosure, the communication path control device 30 establishes a communication path to transmit data from the data source 10 to the server 40. At this time, the data collection system of this disclosure executes the following sequence: The data source 10 requests the allocation of a communication path (S101). The communication path control device 30 establishes a communication path (S102) and then sends a notification that the communication path has been established (S103). The data source 10 transmits data using the communication path established by the communication path control device 30 (S104) and then sends a request to release the communication path (S105). The communication path control device 30 releases the communication path and then sends a notification that the communication path has been released (S106).

[0025] The timing of step S101 in the data transmission sequence is determined by the transmission policy of the data source 10. Examples of transmission policies include: First transmission policy: Establish a communication path and transmit data immediately after data generation. Second transmission policy: Establish a communication path and transmit data after a certain amount of data has accumulated.

[0026] Figure 5 shows an example where the data source 10 transmits data according to the first transmission policy. The time from data generation to data transmission (S104) is the transmission delay ΔS. When many requests arrive at the communication path control device 30, the establishment of the communication path is delayed, and therefore the transmission delay ΔS becomes longer.

[0027] Therefore, in this disclosure, the communication path control device 30 that controls the communication path performs priority control when establishing the communication path. As a result, the data acquisition system of this disclosure can prevent delays in establishing the communication path and prevent delays in data transmission.

[0028] (First Embodiment) Figure 6 shows an example of the configuration of the request receiving unit 31 and the path setting unit 32. The request receiving unit 31 includes queues 312H and 312L and a distribution unit 311. 312H is a request R for assigning a high-priority communication path. Ｓ (H) is a queue for accumulating, and 312L is a request for allocation of a low-priority communication path R. Ｓ (L) is a queue for accumulating. The path setting unit 32 comprises a control unit 321 and a resource management unit 322. In this disclosure, a request to allocate a communication path with a high priority is referred to as a "high priority request," a request to allocate a communication path with a low priority is referred to as a "low priority request," and an allocation request that does not distinguish between these is referred to as R Ｓ It is sometimes referred to as such.

[0029] The distribution unit 311 processes the high-priority request R Ｓ (H) is assigned to queue 312H, and low priority request R Ｓ (L) is allocated to queue 312L. Allocation request R in allocation unit 311. ＳAny method may be used to identify the priority. For example, the priority can be described in a specific field of the packet header, and identification can be performed by reading said priority. However, description in the specific field can also be omitted by, for example, predetermining high priority and low priority for each data source 10 in advance.

[0030] The resource management unit 322 manages resources used by the network device 81. The control unit 321 controls communication paths in the network device 81 by referring to resources managed by the resource management unit 322. In the present embodiment, the resource management unit 322 stores high priority requests R Ｓ high-priority dedicated resources for allocation to (H), and the high-priority requests R Ｓ (H) and low-priority requests R Ｓ (L) are both provided with shared resources that can be used. Thereby, the control unit 321 can provide the high-priority requests R Ｓ (H) with quickly allocated communication paths.

[0031] In a case where such high-priority requests and low-priority requests exist, as shown in FIG. 7, if a high-priority request arrives after a low-priority allocation request is received and all resources have been allocated, the high-priority queue 312H is kept waiting, which increases transmission delay and may cause the delay requirement to no longer be satisfied. In this way, when low-priority requests use shared resources depending on the order of arrival, waiting occurs for high-priority requests, and delay increases despite the high priority of such requests.

[0032] The types of data collected in the data collection system are diverse, and some data should not be delayed, while others may be delayed. Therefore, the path setting unit 32 determines how much leeway is available for each allocation request that arrives at the request receiving unit 31, and delays the establishment of the communication path for those that have sufficient leeway. In this embodiment, in order to determine the leeway for each allocation request, each data source 10 includes information on the amount of data to be transmitted and the delay requirements in the allocation request. However, it is also possible to omit the inclusion of this information in the allocation request by predetermining the amount of data to be transmitted and the delay requirements for each data source 10. The delay requirement information is, for example, the delay time that is allowed for data transmission from the data source 10 to the server 40. This will be explained in detail below.

[0033] Figure 8 shows an example of the resource management method of this embodiment. The resource management method of this embodiment executes steps S31 to S36. In step S31, the distribution unit 311 receives an allocation request, associates the timestamp of when the allocation request arrived with the allocation request, and distributes it to queue 312H or 312L. The time of the timestamp associated with the allocation request at this time is called the "allocation request arrival time".

[0034] The allocation request also includes the amount of data to be transmitted. The control unit 321 uses the amount of data to be transmitted for each allocation request stored in queues 312H and 312L to determine the transmission time T required to transmit the amount of data to be transmitted as specified in the allocation request. Ｓ Calculate this transmission time T. Ｓ This is the time it takes for data sent from the data source to arrive at the server, and can be calculated as link speed / data volume.

[0035] In step S32, the control unit 321 determines the transmission time T Ｓ and grace period T Ｇ Compare the grace period T Ｇ We determine whether the value is greater, i.e., whether there is a grace period. Here, the grace period T Ｇ This can be calculated using the following formula: T Ｇ = T Ｐ - (t-T Ａ) Here, the parameters are as follows: T Ｐ : Allowable delay time in the delay requirement t: Current time T Ａ : The time when the allocation request arrived at the communication path control device 30.

[0036] If the control unit 321 determines in step S32 that there is time to spare, it places the allocation request in queue 312H or 312L, as shown in Figure 9 (S33).

[0037] On the other hand, if the control unit 321 determines in step S32 that there is no grace period for the allocation request in queue 312H, it performs interrupt processing for that allocation request (steps S34 to S36). Specifically, as shown in Figure 10, the control unit 321 temporarily suspends the allocation request with the longest grace period among all allocation requests (S34). At this time, the control unit 321 may release resources from communication paths that have a grace period for delay requirements among the communication paths using shared resources. Next, the control unit 321 returns the suspended allocation request to the front of the queue (S35). Next, the control unit 321 allocates resources to the allocation request that was determined to have no grace period in step S32 (S36).

[0038] As explained above, in this embodiment, when a new allocation request arrives, an interrupt is determined based on whether the delay requirement is met. If the delay requirement is met without interruption, no interrupt is performed; otherwise, an interrupt is performed. When an interrupt is performed, the communication path setting process for the allocation request with the longest grace period among those currently being allocated is stopped and returned to the front of the queue. This makes it possible to achieve data transmission that satisfies the delay requirement for newly arriving high-priority requests in this embodiment.

[0039] (Second Embodiment) Figure 11 shows an example of the resource management method of this embodiment. The resource management method of this embodiment has steps S41 to S43 between steps S31 and S32.

[0040] In step S41, when the control unit 321 retrieves the allocation request from the queue and performs the allocation process, it refers to the resource management unit 322 and determines whether there is a resource equivalent to the amount of data to be transmitted as described in the allocation request. If there is a resource (Yes in S41), the control unit 321 allocates the resource to the allocation request (S44) and calculates the shortest estimated transmission completion time when the data is transmitted (S45). The estimated transmission completion time is managed by the resource management unit 322.

[0041] If no resources are available (No in S41), the control unit 321 refers to the resource management unit 322 and refers to the shortest scheduled transmission completion time among the allocated resources (S42). The control unit 321 then determines whether the shortest scheduled transmission completion time satisfies the delay requirement (S43). If it is determined in step S43 that the delay requirement is met (Yes in S43), the control unit 321 places the allocation request in a queue (S46). On the other hand, if it is determined in step S43 that the delay requirement is not met, the control unit 321 proceeds to step S32.

[0042] Thus, in this embodiment, a determination is made as to whether the delay requirement is met when there are no resources available for allocation. Furthermore, the resource management unit 322 manages the scheduled transmission completion time, and interrupt processing is not performed when a resource becomes available within the time that satisfies the delay requirement, thereby reducing the processing load on the control unit 321.

[0043] (Third Embodiment) In this embodiment, the communication path control device 30 has a request receiving unit 31 that provides a distribution function according to priority. The request receiving unit 31 in this embodiment has a plurality of queues that differ according to priority as queues for communication path allocation requests. In this embodiment, the control unit 32 waits for an arbitrary amount of time before retrieving low-priority requests in order to prioritize high-priority requests.

[0044] Figure 12 shows an example of the configuration of the request receiving unit 31. The request receiving unit 31 includes queues 312H and 312L for communication path allocation requests, a queue 313 for communication path release requests, and a distribution unit 311.

[0045] The distribution unit 311 requests the release of the communication path R. ＲDistribute it to queue 313, high priority request R Ｓ (H) is assigned to queue 312H, and low priority request R Ｓ (L) is allocated to queue 312L.

[0046] In the distribution unit 311, the communication path assignment request R Ｓ and release request R Ｒ The method of identification is arbitrary. For example, a specific field in the packet header could be used to indicate whether it is an allocation request or a release request, and the request could be identified by reading this field.

[0047] Figure 13 shows an example of the communication path establishment flow in the control unit 32. A high-priority request R is sent to queue 312H. Ｓ When (H) is present (Yes in S11), the control unit 32 receives a high-priority request R from queue 312H. Ｓ (H) is retrieved (S12), and a setting process is performed to establish a communication path (S16). A high-priority request R is sent to queue 312H. Ｓ If (H) is not present (No in S11), the control unit 32 waits for a predetermined time (S13) and then sends a high-priority request R to queue 312H again. Ｓ Check if (H) is present (S14). Here, the waiting time in step S13 is arbitrary, for example 5ms can be used as an example. A high-priority request R is placed in queue 312H. Ｓ When (H) is present (Yes in S14), the control unit 32 receives a high-priority request R from queue 312H. Ｓ (H) is retrieved (S12), and a setting process is performed to establish a communication path (S16). A high-priority request R is sent to queue 312H. Ｓ When (H) is not present (No in S14), the control unit 32 receives a low priority request R from queue 312L. Ｓ (L) is taken out (S15), and a process is performed to establish a communication path (S16).

[0048] Furthermore, the control unit 32 receives a request from queue 313 to release the communication path R. Ｒ The timing of reading the data and releasing the communication path is arbitrary. For example, it may be done in parallel with the process of establishing the communication path, or it may be done during the waiting time in step S13.

[0049] As described above, in this embodiment, when the control unit 32 establishes a communication path, it reads in order from the queue 312H with the highest priority. As a result, this embodiment can prioritize the establishment of high-priority communication paths. Furthermore, the data source 10 that issues the assignment request for a high-priority communication path can shorten the path establishment delay. Therefore, this disclosure enables data transmission by RDMA even in the presence of delays in a wide-area network, and enables efficient data transmission to the server 40.

[0050] In this embodiment, an example was shown in which the distribution unit 311 identifies the priority by describing the priority in a specific field of the packet header, but this disclosure is not limited to this. For example, the packet transmission time may be compared with the delay requirement, and packets with the stricter delay requirement may be distributed to queue 312H.

[0051] (Fourth Embodiment) Since the communication path control device 30 manages the communication path, it can obtain the progress of the path establishment process and the remaining processing time until the establishment of the communication path is completed. Therefore, in this embodiment, the progress of the path establishment process is managed, and low priority request R Ｓ During the processing of (L), a high-priority request R Ｓ If (H) is received, the switchback time T for the lower priority communication path will be used. Ｂ The remaining processing time T until the establishment of a lower-priority communication path is complete. Ｒ Compare this with the pruning time T Ｂ If the time is short, the lower priority communication path is switched back to the higher priority request R. Ｓ Process (H).

[0052] Figure 14 shows an example of the communication path establishment flow in the control unit 32. Low priority request R read from queue 312L Ｓ When the communication path establishment process for (L) is being performed, a high-priority request R is sent to queue 312H. Ｓ When (H) is entered, the control unit 32 executes the following process: Switchback time T for the communication path being processed. Ｂ And the remaining processing time T until the establishment of the communication path being processed is complete. Ｒ And, obtain (S21). Then, cutback time T Ｂ and remaining processing time TＲ Compare (S22). Cutback time T Ｂ The remaining processing time T is greater than Ｒ If the time is long (No in S22), a rollback process is performed (S23), and after the process is completed, a high-priority request R is made. Ｓ (H) processing starts (S25). Cutback time T Ｂ The remaining processing time T is greater than Ｒ If the time is short (Yes in S22), processing continues (S24), and after processing is complete, a high-priority request R Ｓ The process of (H) is started (S25).

[0053] As described above, in this embodiment, the control unit 32 is in the process of establishing a low-priority communication path when a high-priority request R enters the high-priority queue 312H. Ｓ Interrupt processing for (H) is performed. This allows this embodiment to prioritize the establishment of higher-priority communication paths.

[0054] (Other Embodiments) The communication path control device 30 of the present invention can also be realized by a computer and a program, and the program can be recorded on a recording medium or provided via a network.

[0055] 10, 10A, 10B, 10C: Data source 20: Transmission control device 30: Communication path control device 40: Server 50: Sensor 31: Request receiving unit 311: Distribution unit 312H, 312L, 313: Queue 32: Control unit 33: Path setting location table 81A, 81B, 81C, 81D, 81E: Network equipment

Claims

1. A data collection system comprising: multiple data sources; a server that collects data from the multiple data sources; and a communication path control device that, when data is transmitted from one of the multiple data sources to the server, establishes a communication path between the one data source and the server in accordance with an allocation request from the one data source or the server, wherein the communication path control device releases resources from communication paths that have sufficient delay requirements among already established communication paths, and allocates the released resources to high-priority allocation requests.

2. The data collection system according to claim 1, wherein when the communication path control device receives a high-priority allocation request that does not meet the delay requirement, it releases the resources of an already established communication path that has a grace period for the delay requirement.

3. The data collection system according to claim 1, wherein the communication path control device comprises a resource management unit that manages high-priority dedicated resources that are allocated to high-priority allocation requests and shared resources that can be allocated to low-priority allocation requests, and releases resources of communication paths that have a grace period for delay requirements among the communication paths using the shared resources.

4. A resource management method performed by a data collection system comprising: multiple data sources; a server for collecting data from the multiple data sources; and a communication path control device for establishing communication paths between the multiple data sources and the server, wherein the communication path control device releases resources from communication paths that have sufficient delay requirements among already established communication paths, and allocates the released resources to communication paths with higher priority.

Images