I / O request scheduling method, device, electronic device and storage medium
By using a dynamically programmed I/O request scheduling strategy, the resource utilization of the buffer system is optimized, the inefficiency problem when multiple applications access the system concurrently is solved, and the system performance is improved.
Patent Information
- Application Number
- CN202211223430.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-10-08
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2042-10-08
AI Technical Summary
When multiple applications concurrently access I/O buffer resources in a computing system, the existing random scheduling method results in low operating efficiency and resource utilization of the buffer system.
By obtaining the I/O requests to be scheduled and the available cache capacity of the reference cache node, a target scheduling strategy is generated using dynamic programming that minimizes scheduling costs, thereby optimizing the scheduling of I/O requests.
It improves the operating efficiency of the buffer system and the utilization rate of buffer resources, and reduces application latency and interruption probability.
Smart Images

Figure CN115437794B_ABST
Abstract
Description
Background Art
[0002] In the field of computer technology, computing systems typically use burst buffers to cope with peak I / O (Input / Output) requests from external storage. Multiple concurrently running applications share the I / O buffer resources of the burst buffer. However, when I / O requests from multiple applications access the computing system, they compete for the I / O buffer resources, resulting in reduced performance of the computing system.
[0003] In the related art, concurrently accessed I / O requests are randomly scheduled to different cache nodes of the burst buffer. However, the random scheduling of concurrently accessed I / O requests is difficult to adapt to the dynamic changes of the burst buffer, resulting in low operating efficiency of the buffer system and low utilization of buffer resources.
[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention
[0005] The present disclosure provides an I / O request scheduling method, device, electronic device and storage medium, which at least to a certain extent overcome the problems of low operating efficiency of the buffer system and low utilization rate of buffer resources in the related art.
[0006] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.
[0007] According to one aspect of the present disclosure, there is provided an input / output (I / O) request scheduling method, comprising: obtaining a plurality of I / O requests to be scheduled and corresponding data volumes; obtaining available cache capacities of a plurality of reference cache nodes; generating a target scheduling strategy corresponding to the data volumes and the available cache capacities based on dynamic programming that minimizes scheduling costs; and scheduling the plurality of I / O requests to be scheduled to the plurality of reference cache nodes according to the target scheduling strategy.
[0008] According to another aspect of the present disclosure, there is provided an input / output (I / O) request scheduling device, comprising: an acquisition module, used to acquire a plurality of I / O requests to be scheduled and the corresponding data volumes; the acquisition module, further used to acquire the available cache capacity of a plurality of reference cache nodes; a generation module, used to generate a target scheduling strategy corresponding to the data volume and the available cache capacity based on dynamic programming that minimizes scheduling costs; and a scheduling module, used to schedule the plurality of I / O requests to be scheduled to the plurality of reference cache nodes according to the target scheduling strategy.
[0009] In one embodiment of the present disclosure, the acquisition module is used to obtain job information of a first application, where the first application is an application that has completed registration with a scheduling device, and the job information includes a first I / O request of the first application and a corresponding data volume; determine a second application in the first application whose job phase is an I / O phase; determine a second I / O request in the first I / O request that causes the second application to be in the I / O phase; and determine multiple I / O requests to be scheduled and corresponding data volumes based on the second I / O request and the job information.
[0010] In one embodiment of the present disclosure, the acquisition module is used to obtain available cache capacities corresponding to multiple cache nodes to be selected, and the multiple cache nodes to be selected are cache nodes included in a buffer device for receiving I / O requests; and determine that the cache nodes whose available cache capacities are greater than a capacity threshold among the multiple cache nodes to be selected are reference cache nodes.
[0011] In one embodiment of the present disclosure, the acquisition module is used to receive an indication message sent by a buffer device, the indication message includes available cache capacity corresponding to the multiple reference cache nodes, and the available cache capacity corresponding to the multiple reference cache nodes is greater than a capacity threshold; based on the indication message, the available cache capacity of the reference cache node is determined.
[0012] In one embodiment of the present disclosure, the generation module is used to determine the data volume and the available cache capacity, and one or more intermediate scheduling costs corresponding to the data volume and the available cache capacity under one or more dynamic programming methods for minimizing the scheduling cost; determine the minimum scheduling cost among the one or more intermediate scheduling costs; and determine that the scheduling strategy corresponding to the minimum scheduling cost is the target scheduling strategy.
[0013] In one embodiment of the present disclosure, the number of the multiple I / O requests to be scheduled is n, the number of the multiple reference cache nodes is m, and n and m are both integers greater than or equal to 2; the one or more dynamic programming methods for minimizing the scheduling cost include one or more of the following dynamic programming methods:
[0014] C ij +OPT(n -i ,m -j )
[0015] OPT -i ,m)
[0016] OPT(n,m -j )
[0017] OPT -i ,m -j )
[0018] Among them, C ij is the scheduling cost of scheduling the i-th I / O request to be scheduled to the j-th reference cache node; OPT(n -i ,m -j ) is the minimum scheduling cost of scheduling the other I / O requests to be scheduled except the i-th I / O request to be scheduled among the n I / O requests to be scheduled to the other cache nodes except the j-th reference cache node among the m reference cache nodes based on dynamic programming; OPT(n -i OPT(n,m) is the minimum scheduling cost of scheduling the n I / O requests to be scheduled except the i-th I / O request to be scheduled to m reference cache nodes based on dynamic programming; OPT(n,m -j ) is based on dynamic programming, and is a method for scheduling n I / O requests to be scheduled to other reference cache nodes among m reference cache nodes except the j-th reference cache node.
[0019] In one embodiment of the present disclosure, when the purpose of the target scheduling policy is to maximize the bandwidth obtained by the application in the buffer device, the C ij The duration is determined based on the number of the multiple I / O requests to be scheduled, the corresponding data volume, and the completion time; the completion time is the time taken by the multiple reference cache nodes to complete the corresponding I / O requests to be scheduled.
[0020] In one embodiment of the present disclosure, when the objective of the target scheduling strategy is to maximize the buffer system efficiency, the C ij according to and At least one of the following is determined: is the average value of the sum of the ratios of each service duration to the previous service duration in a plurality of service durations, wherein the plurality of service durations are the durations during which the plurality of reference cache nodes provide services for a plurality of third applications; the plurality of third applications are applications corresponding to the plurality of I / O requests to be scheduled; is the average remaining cache capacity of the multiple reference cache nodes.
[0021] In one embodiment of the present disclosure, when the objective of the target scheduling strategy is to maximize the fairness of the application program, the C ijDetermined according to the first completion time and the second completion time; the first completion time is the time when the multiple reference cache nodes complete the service for the first third application; the second completion time is the time when the multiple reference cache nodes complete the service for the last third application; wherein, the first third application is the first third application registered with the scheduling device among the multiple third applications; the last third application is the last third application registered with the scheduling device among the multiple third applications.
[0022] In one embodiment of the present disclosure, when the purpose of the target scheduling strategy is to minimize the delay duration of the application program, the C ij The waiting scheduling time and the additional time corresponding to each I / O request to be scheduled are determined according to the multiple I / O requests to be scheduled; the waiting scheduling time is the difference between the time point when the multiple reference cache nodes serve the corresponding I / O request to be scheduled and the time point when they serve the first I / O request to be scheduled among the multiple I / O requests to be scheduled; the additional time is the difference between the time when the multiple reference cache nodes serve the corresponding I / O request to be scheduled and the time when the multiple reference cache nodes serve the corresponding I / O request to be scheduled in a non-interference situation.
[0023] In one embodiment of the present disclosure, the acquisition module is also used to obtain the concurrency tolerance of the cache node to be detected, wherein the concurrency tolerance represents the concurrency capability of the cache node to be detected; obtain the concurrent number of applications concurrently accessed in the cache node to be detected; obtain the achievable maximum bandwidth and actual bandwidth of the cache node to be detected; the device also includes: a first determination module, which is used to determine the sensitivity of the cache node to be detected to concurrent access based on the concurrency tolerance, the concurrent number, the achievable maximum bandwidth and the actual bandwidth.
[0024] In one embodiment of the present disclosure, the concurrency tolerance includes one or more of the bus channels, controller channels, and the number of storage bodies corresponding to the cache node to be detected.
[0025] In one embodiment of the present disclosure, the first determination module is used to calculate the sensitivity according to the following formula:
[0026]
[0027]
[0028] δ=λ τ ×λ ω
[0029] Among them, τ ais the number of concurrent connections; τ dc is the concurrency tolerance; ω max is the maximum achievable bandwidth; ω r is the actual bandwidth; δ is the sensitivity.
[0030] In one embodiment of the present disclosure, the acquisition module is also used to obtain a first execution duration of a job corresponding to the application to be detected under interference conditions, wherein the interference indicates that there are other applications executed in parallel in the buffer device; and to obtain a second execution duration of the job under no interference conditions, wherein the no interference indicates that there are no other applications executed in parallel in the buffer device; the device also includes: a second determination module, used to determine an interference value based on the first execution duration and the second execution duration.
[0031] In one embodiment of the present disclosure, the second determination module is used to calculate the interference value according to the following formula:
[0032]
[0033] Among them, ET with_interference The first execution time; ET without_interference is the second execution duration; and ε is the interference value.
[0034] According to another aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute any of the above-mentioned input / output (I / O) request scheduling methods by executing the executable instructions.
[0035] According to another aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, any of the above-mentioned input / output (I / O) request scheduling methods is implemented.
[0036] According to another aspect of the present disclosure, a computer program product is provided, which includes a computer program or computer instructions, and the computer program or the computer instructions are loaded and executed by a processor to enable a computer to implement any of the above-mentioned input / output I / O request scheduling methods.
[0037] The technical solution provided by the embodiments of the present disclosure includes at least the following beneficial effects:
[0038] The technical solution provided by the embodiments of the present disclosure obtains the data volume of multiple I / O requests to be scheduled and the available cache capacity of multiple reference cache nodes, and uses the data volume and available cache capacity to perform dynamic programming to minimize the scheduling cost, and generates a target scheduling strategy for scheduling multiple I / O requests to be scheduled. The target scheduling strategy generated using the data volume of the I / O requests to be scheduled and the available cache capacity of the reference cache node takes into account the available cache capacity representing the state of the reference cache node and the data volume representing the state of the I / O requests to be scheduled, so that the generated target scheduling strategy can better adapt to the dynamic changes of the buffer where the reference cache node is located. In addition, executing the target scheduling strategy generated based on dynamic programming to minimize the scheduling cost can improve the operating efficiency of the buffer system and the utilization rate of buffer resources.
[0039] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.
[0041] Figure 1 A schematic diagram showing the structure of an I / O request scheduling system in an embodiment of the present disclosure is shown;
[0042] Figure 2 A flow chart of an I / O request scheduling method in one embodiment of the present disclosure is shown;
[0043] Figure 3 A flow chart of determining the sensitivity of a cache node to be detected to concurrent access in one embodiment of the present disclosure is shown;
[0044] Figure 4 A flow chart of a method for determining an interference value of an application program in one embodiment of the present disclosure is shown;
[0045] Figure 5 A flow chart of an I / O request scheduling method in another embodiment of the present disclosure is shown;
[0046] Figure 6 A schematic diagram of an I / O request scheduling device in an embodiment of the present disclosure is shown;
[0047] Figure 7 A structural block diagram of an electronic device in an embodiment of the present disclosure is shown. DETAILED DESCRIPTION
[0048] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the disclosure will be more comprehensive and complete and to fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
[0049] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.
[0050] Figure 1 A schematic diagram showing the structure of an I / O request scheduling system in an embodiment of the present disclosure is shown. The system can apply the I / O request scheduling method or I / O request scheduling device in various embodiments of the present disclosure.
[0051] like Figure 1 As shown, the I / O request scheduling system architecture may include: an I / O scheduling device 11 , a buffer device 12 , and multiple access devices 13 .
[0052] The buffer device 12 includes a plurality of cache nodes, each of which has certain cache resources, which can be used to cache I / O requests, etc. The buffer device 12 is configured with a monitoring module, which can monitor the status information of the plurality of cache nodes, for example, the available cache capacity, bandwidth, busy / idle information of the plurality of cache nodes, etc. The buffer device 12 can send data to the I / O scheduling device 11, and the buffer device 12 can also receive data sent by the I / O scheduling device 11.
[0053] An application is installed in each device included in the multiple access devices 13. The application installed in the multiple access devices 13 generates a corresponding job during startup and operation, and the job includes an I / O request. The access devices included in the multiple access devices 13 can send the I / O request generated by the application to the I / O scheduling device 11. In some embodiments, the access device corresponding to the application can register the information of the job generated by the application with the I / O scheduling device 11, for example, the information includes the name, size, content, etc. of the job.
[0054] In some embodiments, an access device corresponding to an application is configured with an agent module, which can monitor the current stage of the application. The stage of the application is the stage of the job being performed by the application. For example, a job usually includes an I / O request and a computing task. When the application needs to read / write data, the current job stage is the I / O stage, and accordingly, the stage of the application is the I / O stage. In some embodiments, the agent module can report the stage information of the monitored application to the corresponding access device, and the access device can send the stage information to the I / O scheduling device 11.
[0055] The I / O scheduling device 11 maintains an I / O scheduling queue. In one embodiment, the I / O queue includes scheduled but unexecuted I / O requests and unscheduled I / O requests. In another embodiment, the I / O queue includes unscheduled I / O requests. Among them, the I / O scheduling device 11 can receive the stage information of the application sent by the proxy module, determine the I / O requests that need to be scheduled based on the stage information and the job information of the application, and then maintain the I / O scheduling queue based on the I / O requests that need to be scheduled. The I / O scheduling device 11 can receive the node status information sent by the buffer device 12, and maintain the status of the cache node in the buffer device 11 based on the status information. The I / O scheduling device 11 can generate a corresponding scheduling strategy based on the I / O requests that need to be scheduled in the scheduling queue and the status of the cache nodes included in the buffer device 12, and schedule the I / O requests to the corresponding cache nodes according to the scheduling strategy.
[0056] The I / O scheduling device 11, the buffer device 12, and the multiple access devices 13 are connected to each other through a network, and the network can be a wired network or a wireless network.
[0057] Optionally, the wireless network or wired network described above uses standard communication technology and / or protocol. The network is usually the Internet, but it can also be any network, including but not limited to a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a mobile, wired or wireless network, a dedicated network or any combination of a virtual private network). In some embodiments, technologies and / or formats including Hyper Text Mark-up Language (HTML), Extensible Markup Language (XML), etc. are used to represent data exchanged through the network. In addition, conventional encryption technologies such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), Internet Protocol Security (IPsec) can also be used to encrypt all or some links. In other embodiments, customized and / or dedicated data communication technologies can also be used to replace or supplement the above data communication technologies.
[0058] The access devices included in the multiple access devices 13 may be various electronic devices, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, wearable devices, augmented reality devices, virtual reality devices, etc. The applications installed in different access devices included in the multiple access devices 13 may be the same application or different application, which is not limited in the embodiments of the present disclosure.
[0059] The buffer device 12 is an electronic device having one or more cache nodes. For example, the buffer device 12 may be a server or terminal with storage function. The cache node may be an independent electronic device, and the storage media corresponding to different cache nodes may be the same or different. For example, the storage medium corresponding to each cache node may be one or more of RAM (Random Access Memory), SSD (Solid State Disk), and HDD (Hard Disk Drive).
[0060] The I / O scheduling device 11 may be any electronic device having the above functions, such as a server or a terminal. In one embodiment, the I / O scheduling device 11 may also be a functional module in the buffer device 12 .
[0061] Those skilled in the art will know that Figure 1 The number of the multiple access devices 13 is only illustrative, and any number of access devices may be provided according to actual needs. This embodiment of the present disclosure does not limit this.
[0062] The present exemplary implementation is described in detail below with reference to the accompanying drawings and embodiments.
[0063] The present disclosure provides an I / O request scheduling method, which can be executed by any electronic device with computing and processing capabilities. For example, the electronic device is an I / O scheduling device.
[0064] Figure 2 A flow chart of an I / O request scheduling method according to an embodiment of the present disclosure is shown as follows: Figure 2 As shown, the I / O request scheduling method provided in the embodiment of the present disclosure includes the following S201 to S204.
[0065] S201, obtaining multiple I / O requests to be scheduled and corresponding data volumes.
[0066] In some embodiments, the I / O requests to be scheduled include unscheduled I / O requests. In another embodiment, the I / O requests to be scheduled include unscheduled I / O requests and scheduled but unexecuted I / O requests.
[0067] In some embodiments, obtaining multiple I / O requests to be scheduled and the corresponding data volume may include: obtaining job information of a first application, the first application is an application that has completed registration with a scheduling device, and the job information includes a first I / O request of the first application and the corresponding data volume; determining a second application whose job phase is the I / O phase in the first application; determining a second I / O request in the first I / O request that puts the second application in the I / O phase; and determining multiple I / O requests to be scheduled and the corresponding data volume according to the second I / O request and the job information. The job information includes the name, size, content, and data volume and startup time of the job corresponding to the first application, and the data volume corresponding to each task phase includes the data volume corresponding to the I / O request in the job information.
[0068] In some embodiments, when an application is started, the access device where the application is located registers the job information corresponding to the application with the I / O scheduling device. That is, obtaining the job information of the first application may include: receiving the job information sent by the access device where the first application is located.
[0069] In some embodiments, the device where the application is located monitors the operating stage of the application, obtains the operating stage information of the application, and then sends the operating stage information to the I / O scheduling device. Determining the second application whose operating stage is the I / O stage in the first application may include: receiving the operating stage information corresponding to the first application; and determining the second application whose operating stage is the I / O stage in the first application based on the operating stage information corresponding to the first application. The I / O scheduling device can determine the operating stage of the application in real time based on the operating stage information, and accordingly, can directly determine the second I / O request that puts the second application in the I / O stage from the first I / O request included in the operating information.
[0070] The I / O scheduling device maintains an I / O scheduling queue, which has information about I / O requests that have been scheduled but not executed in the I / O scheduling queue, and can determine multiple I / O requests to be scheduled and corresponding data volumes based on the second I / O scheduling request and the scheduled but not executed I / O requests.
[0071] S202: Obtain available cache capacities of multiple reference cache nodes.
[0072] In some embodiments, obtaining available cache capacity of multiple reference cache nodes may include: obtaining available cache capacity corresponding to multiple cache nodes to be selected, where the multiple cache nodes to be selected are cache nodes included in a buffer device for receiving I / O requests; determining a cache node among the multiple cache nodes to be selected whose available cache capacity is greater than a capacity threshold as a reference cache node.
[0073] The buffer device is configured with a detection module, which can monitor the available buffer capacity of multiple cache nodes to be selected included in the buffer device, and can send the available buffer capacity information of multiple cache nodes to be selected to the I / O scheduling device. After receiving the available buffer capacity information, the I / O scheduling device can determine the available buffer capacity of multiple cache nodes to be selected based on the available buffer capacity information, and determine the reference cache node whose available buffer capacity is greater than the capacity threshold among the multiple cache nodes to be selected based on the available buffer capacity of the multiple cache nodes to be selected. Among them, the capacity threshold is a set value. Regarding the specific size of the capacity threshold, the embodiment of the present disclosure does not limit it and can be set based on experience.
[0074] In other embodiments, obtaining the available cache capacity of multiple reference cache nodes may include: receiving an indication message sent by a buffer device, the indication message including the available cache capacity corresponding to the multiple reference cache nodes, the available cache capacity corresponding to the multiple reference cache nodes being greater than a capacity threshold; and determining the available cache capacity of the reference cache node based on the indication message.
[0075] In some embodiments, the buffer device is configured with a monitoring module, which can monitor the available buffer capacity of multiple cache nodes to be selected included in the buffer device, and when the available cache capacity of the cache node to be selected is greater than the capacity threshold, determine that the cache node to be selected is a reference cache node, and generate an indication message, which includes the identifier of the reference cache node among the multiple cache nodes to be selected and the corresponding available cache capacity.
[0076] S203, generating a target scheduling strategy corresponding to the data volume and available cache capacity based on dynamic programming that minimizes the scheduling cost.
[0077] In some embodiments, generating a target scheduling strategy corresponding to the data volume and available cache capacity based on dynamic programming that minimizes scheduling costs may include: determining one or more intermediate scheduling costs corresponding to the data volume and available cache capacity under one or more dynamic programming methods that minimize scheduling costs; determining the minimum scheduling cost among the one or more intermediate scheduling costs; and determining the scheduling strategy corresponding to the minimum scheduling cost as the target scheduling strategy.
[0078] In some embodiments, the number of the plurality of I / O requests to be scheduled is n, the number of the plurality of reference cache nodes is m, n and m are both integers greater than or equal to 2, and the one or more dynamic programming methods for minimizing the scheduling cost include one or more of the following dynamic programming methods:
[0079] C ij +OPT(n -i ,m -j ) (1)
[0080] OPT -i ,m) (2)
[0081] OPT(n,m -j ) (3)
[0082] OPT -i ,m -j ) (4)
[0083] Among them, C ij is the scheduling cost of scheduling the i-th I / O request to be scheduled to the j-th reference cache node. OPT(n -i ,m -j ) is the minimum scheduling cost of scheduling the other I / O requests to be scheduled among the n I / O requests to be scheduled except the i-th I / O request to be scheduled to the other cache nodes among the m reference cache nodes except the j-th reference cache node based on dynamic programming. OPT(n -iOPT(n,m) is the minimum scheduling cost of scheduling the n I / O requests to be scheduled except the i-th I / O request to be scheduled to m reference cache nodes based on dynamic programming. -j ) is based on dynamic programming, and is a method for scheduling n I / O requests to be scheduled to other reference cache nodes among m reference cache nodes except the j-th reference cache node.
[0084] In some embodiments, the minimum scheduling cost may be determined according to the dynamic programming shown in the following formula 5, and the scheduling strategy corresponding to the minimum scheduling cost may be determined as the target scheduling strategy.
[0085]
[0086] Among them, OPT(n,m) is the minimum scheduling cost of scheduling n I / O requests to be scheduled to m reference cache nodes under the four dynamic programming methods shown in Formula 1, Formula 2, Formula 3 and Formula 4, and MIN(.) is the minimum value function.
[0087] In some embodiments, when the target scheduling policy corresponds to maximizing the bandwidth obtained by the application in the buffer device, C ij Determined according to the number of multiple I / O requests to be scheduled and the corresponding data volume and completion time; the completion time is the time taken by multiple reference cache nodes to complete the corresponding I / O requests to be scheduled. In some embodiments, the completion time can be predicted based on the data volume of multiple I / O requests to be scheduled and the available cache capacity of multiple reference cache nodes. The embodiments of the present disclosure do not limit how to predict the completion time of the I / O requests to be scheduled based on the data volume of multiple I / O requests to be scheduled and the available cache capacity of multiple reference cache nodes. For example, the prediction can be made by direct calculation, or the completion time can be predicted by an empirical model. Among them, the empirical model generates the completion time of the current I / O request to be scheduled based on the completion time of the I / O requests to be scheduled in the historical data of the scheduling under similar circumstances to the current scheduling.
[0088] In one embodiment, when the target scheduling policy corresponds to maximizing the bandwidth obtained by the application in the buffer device, C ij It is obtained according to the following formula 6.
[0089]
[0090] Among them, γ i is the amount of data for the ith I / O request to be scheduled, t i is the completion time of the i-th I / O request to be scheduled, and n is the number of multiple I / O requests to be scheduled, where n is an integer greater than or equal to 2.
[0091] In some embodiments, when the target scheduling strategy corresponds to the maximum buffer system efficiency, C ij according to and At least one of is determined. It is the average value of the sum of the ratios of each service duration to the previous service duration in the multiple service durations, and the multiple service durations are the durations during which the multiple reference cache nodes provide services to the multiple third applications. The multiple third applications are the applications corresponding to the multiple I / O requests to be scheduled. is the average remaining cache capacity of multiple reference cache nodes.
[0092] The registration times of multiple third applications with the I / O scheduling device have a sequence relationship. Based on this sequence relationship, the order in which the third applications start to be served can be determined. Based on the sequence, the service duration of each third application served by multiple reference cache nodes and the ratio of each service duration to the previous service duration can be determined. For the first third application served by multiple reference cache nodes, there is no corresponding previous third application served. Therefore, when calculating the service duration, the service duration corresponding to the first third application can be used only as the denominator of the subsequent calculation of the service duration ratio. For example, there are 3 third applications in total, and the service duration corresponding to the first third application is 1s (seconds), the service duration corresponding to the second third application is 1.1s, and the service duration corresponding to the third third application is 2.2s.
[0093] In some embodiments, the service duration corresponding to each third application can be predicted based on the data volume of multiple I / O requests to be scheduled and the available cache capacity of multiple reference cache nodes. The embodiments of the present disclosure do not limit how to predict the service duration corresponding to each third application based on the data volume of multiple I / O requests to be scheduled and the available cache capacity of multiple reference cache nodes. For example, the prediction can be made by direct calculation, or the service duration can be predicted by an empirical model.
[0094] In some embodiments, It can be obtained by adding up the available cache capacities of multiple reference cache nodes and taking the average value after each I / O request to be scheduled is allocated a corresponding cache resource. The disclosed embodiments do not limit how to predict the available cache capacity of multiple reference cache nodes based on the data volume of multiple I / O requests to be scheduled and the available cache capacity of multiple reference cache nodes. For example, the prediction can be made by direct calculation, or the available cache capacity can be predicted by an empirical model. For example, the available cache capacity of multiple reference cache nodes after each I / O request to be scheduled is allocated a corresponding cache resource can be directly calculated based on the data volume of multiple I / O requests to be scheduled and the available cache capacity of multiple reference cache nodes before cache resources are allocated to the I / O requests to be scheduled.
[0095] In some embodiments, when the target scheduling strategy corresponds to the maximum buffer system efficiency, C ij It is obtained according to the following formula 7.
[0096]
[0097] In some embodiments, when the objective of the target scheduling policy is to maximize the fairness of the application program, C ij Determined based on the first completion time and the second completion time; the first completion time is the time when multiple reference cache nodes complete services for the first third application; the second completion time is the time when multiple reference cache nodes complete services for the last third application; wherein, the first third application is the first third application registered with the scheduling device among multiple third applications; the last third application is the last third application registered with the scheduling device among multiple third applications.
[0098] The first completion time and the second completion time can be predicted based on the data volume of multiple I / O requests to be scheduled and the available cache capacity of multiple reference cache nodes. The embodiments of the present disclosure do not limit how to predict the first completion time and the second completion time based on the data volume of multiple I / O requests to be scheduled and the available cache capacity of multiple reference cache nodes. For example, the prediction can be made by direct calculation, or the first completion time and the second completion time can be predicted by an empirical model.
[0099] In some embodiments, when the objective of the target scheduling policy is to maximize the fairness of the application program, C ij It is obtained according to the following formula 8.
[0100] C ij =t c-LastApp -t c-FirstApp (8)
[0101] Among them, tc-LastApp is the second completion time, t c-FirstApp The first completion time.
[0102] In some embodiments, when the purpose of the target scheduling policy is to minimize the delay duration of the application program, C ij The waiting scheduling time and the additional time corresponding to each I / O request to be scheduled are determined according to multiple I / O requests to be scheduled; the waiting scheduling time is the difference between the time point when multiple reference cache nodes serve the corresponding I / O request to be scheduled and the time point when they serve the first I / O request to be scheduled among the multiple I / O requests to be scheduled; the additional time is the difference between the time when multiple reference cache nodes serve the corresponding I / O request to be scheduled and the time when multiple reference cache nodes serve the corresponding I / O request to be scheduled in a non-interference situation.
[0103] Among them, in the absence of interference, the duration information of multiple reference cache nodes serving the corresponding I / O requests to be scheduled can be stored in advance in the I / O scheduling device. For example, based on the historical data of the scheduled I / O requests, it is possible to determine the duration of multiple reference cache nodes serving the historical I / O requests to be scheduled, and this duration can be considered as the duration of the current I / O request to be scheduled being served without interference. The historical I / O request to be scheduled has the same amount of data as the current I / O request to be scheduled, the environment is the same or similar, and the historical I / O request to be scheduled is not interfered by other I / O requests when it is executed. Among them, the same or similar environment includes that the multiple cache nodes included in the buffer device have the same or similar available cache capacity.
[0104] In some embodiments, the length of time that each I / O request to be scheduled waits for scheduling, and the length of time that multiple reference cache nodes serve the corresponding I / O requests to be scheduled, can be predicted based on the data volume of the multiple I / O requests to be scheduled and the available cache capacity of the multiple reference cache nodes.
[0105] In some embodiments, when the purpose of the target scheduling policy is to minimize the delay duration of the application program, C ij It is obtained according to the following formula 9.
[0106]
[0107] Where n is the number of I / O requests to be scheduled, t i-waiting is the waiting time for the i-th I / O request to be scheduled, t i-extra It is the additional time taken by the i-th I / O request to be scheduled.
[0108] S204: Schedule multiple I / O requests to be scheduled to multiple reference cache nodes according to the target scheduling policy.
[0109] The I / O scheduling device generates a target scheduling policy based on the data volume of multiple I / O requests to be scheduled and the available cache capacity of multiple reference cache nodes, and then executes the target scheduling policy, allocates cache resources indicated in the target scheduling policy to the multiple I / O requests to be scheduled from the buffer resources of the multiple reference cache nodes, and schedules the multiple I / O requests to be scheduled to the multiple reference cache nodes.
[0110] The technical solution provided by the embodiments of the present disclosure obtains the data volume of multiple I / O requests to be scheduled and the available cache capacity of multiple reference cache nodes, and uses the data volume and available cache capacity to perform dynamic programming to minimize the scheduling cost, and generates a target scheduling strategy for scheduling multiple I / O requests to be scheduled. The target scheduling strategy generated using the data volume of the I / O requests to be scheduled and the available cache capacity of the reference cache node takes into account the available cache capacity representing the state of the reference cache node and the data volume representing the state of the I / O requests to be scheduled, so that the generated target scheduling strategy can better adapt to the dynamic changes of the buffer where the reference cache node is located. In addition, executing the target scheduling strategy generated based on dynamic programming that minimizes the scheduling cost can improve the operating efficiency of the buffer system and the utilization rate of buffer resources, thereby reducing the probability of significant delays or interruptions in the operation of the application corresponding to the I / O requests to be scheduled.
[0111] In addition, the embodiment of the present disclosure also provides a scheduling strategy for maximizing the bandwidth obtained by the application in the buffer device, or for maximizing the buffer system efficiency, or for maximizing the fairness of the application, or for minimizing the delay time of the application. ij The calculation method of .
[0112] In another embodiment of the present disclosure, a method for determining the sensitivity of a cache node to be detected to concurrent access is also provided. Figure 3 As shown, the process of determining the sensitivity of the cache node to be detected to concurrent access includes S301 to S304.
[0113] S301, obtaining the concurrency tolerance of the cache node to be detected, where the concurrency tolerance indicates the concurrency capability of the cache node to be detected.
[0114] The cache nodes to be detected are some or all of the cache nodes included in the buffer device. For example, the buffer device includes 5 cache nodes, and the cache nodes to be detected can be any 3 of the 5 cache nodes, or can be all of the 5 cache nodes. In some embodiments, the concurrency tolerance of the cache nodes to be detected includes one or more of the bus channels, controller channels, and storage volume numbers corresponding to the cache nodes to be detected.
[0115] The bus channels, controller channels, and number of storage bodies corresponding to the cache node to be detected are all determined information, which can be stored in the buffer device. Obtaining the concurrency tolerance of the cache node to be detected can include: sending a request message to the buffer device; receiving a feedback message sent by the buffer device, and the feedback message includes one or more of the bus channels, controller channels, and number of storage bodies corresponding to the cache node to be detected.
[0116] S302, obtaining the concurrent number of applications concurrently accessed in the cache node to be detected.
[0117] The concurrency number is used to indicate the number of concurrently accessed applications in the cache node to be detected. The I / O scheduling device can schedule the I / O requests to be scheduled. Accordingly, the I / O scheduling device stores the correspondence between the I / O requests and the cache nodes. Based on the correspondence, the concurrency number of concurrently accessed applications in the cache node to be detected can be obtained.
[0118] S303: Obtain the achievable maximum bandwidth and actual bandwidth of the cache node to be detected.
[0119] The achievable maximum bandwidth and actual bandwidth of the cache node to be detected are determined information. In some embodiments, the achievable maximum bandwidth and actual bandwidth of the cache node to be detected can be stored in a buffer device. Obtaining the achievable maximum bandwidth and actual bandwidth of the cache node to be detected can include: obtaining the achievable maximum bandwidth and actual bandwidth of the cache node to be detected from the buffer device.
[0120] S304: Determine the sensitivity of the cache node to be detected to concurrent access according to the concurrency tolerance, the number of concurrent accesses, the maximum achievable bandwidth, and the actual bandwidth.
[0121] In some embodiments, the sensitivity of the cache node to be detected to concurrent access is determined based on concurrency tolerance, concurrency number, maximum achievable bandwidth and actual bandwidth, and the sensitivity of the cache node to be detected to concurrent access can be calculated according to the following formulas 10, 11 and 12.
[0122]
[0123]
[0124] δ=λ τ ×λ ω (12)
[0125] Among them, τ a is the number of concurrent connections; τ dc is the concurrency tolerance; ω max To achieve the maximum bandwidth; ris the actual bandwidth; δ is the sensitivity of the cache node to be detected to concurrent access.
[0126] The technical solution provided by the embodiment of the present disclosure provides a method for expressing the sensitivity of a cache node to concurrent access, and provides a method for calculating the sensitivity of a cache node to be detected to concurrent access, thereby providing a reference for evaluating the performance of cache nodes and the effects of different scheduling strategies.
[0127] In another embodiment of the present disclosure, a method for determining an interference value of an application is also provided. Figure 4 As shown, the process of determining the interference value of the application includes S401 to S403.
[0128] S401, obtaining a first execution duration of a job corresponding to the application to be detected in a condition of interference, where interference means that other applications are executed in parallel in the buffer device.
[0129] The application to be detected can be any application that requests cache resources from the buffer device. After the application to be detected is started, the access device corresponding to the application to be detected registers the job information of the application to be detected with the I / O scheduling device, and records the completion time of the job when the application completes the job, and then sends the job completion time to the I / O scheduling device. Based on the registration time and the job completion time, the I / O scheduling device can obtain the first execution time of the job corresponding to the application to be detected under interference.
[0130] S402, obtaining a second execution duration of the job in a non-interference situation, where non-interference means that there are no other application programs executed in parallel in the buffer device.
[0131] In some embodiments, in the absence of interference, the second execution duration of the job of the application to be detected can be the execution duration of a historical job as the second execution duration, the historical job has the same amount of data as the job of the application to be detected, and the environment in which the historical job is executed is the same or similar to the environment in which the job of the application to be detected is executed, for example, the cache nodes included in the buffer device have the same or similar available cache capacity, and there is no interference from the jobs of other applications when executing the historical job. Obtaining the second execution duration of the job in the absence of interference may include: obtaining the execution duration of the historical job.
[0132] S403: Determine an interference value according to the first execution duration and the second execution duration.
[0133] In some embodiments, the interference value may be determined according to the first execution duration and the second execution duration as shown in Formula 13.
[0134]
[0135] Among them, ET with_interference The first execution time; ET without_interference is the second execution time; ε is the interference value.
[0136] The technical solution provided by the embodiment of the present disclosure provides a method for representing the interference situation of an application program and a method for calculating the interference degree of an application program, thereby providing a reference for evaluating the performance of cache nodes and the effects of different scheduling strategies.
[0137] For ease of understanding, the following will be summarized Figure 2 The corresponding embodiments, Figure 3 The corresponding embodiments and Figure 4 The corresponding embodiments are used to illustrate the I / O request scheduling method provided by the embodiments of the present disclosure. Figure 5 As shown, the I / O request scheduling method provided in another embodiment of the present disclosure includes S501 to S508.
[0138] S501, the buffer device continuously sends the status of the cache node to the I / O scheduling device. The monitoring module configured in the buffer device continuously collects the status of the cache nodes included in the buffer device, and sends the busy / idle status information of the cache nodes to the I / O scheduling device.
[0139] S502, the buffer device sends an indication message to the I / O scheduling device, the indication message including the identifier of the reference cache node and the corresponding available cache capacity. The available cache capacity of the reference cache node is greater than the capacity threshold. The monitoring module configured in the buffer device continuously collects the available cache capacity of the cache node included in the buffer device, and reports the identifier and available cache capacity of the cache node to the I / O scheduling device when the available cache capacity of the cache node is greater than the capacity threshold.
[0140] S503, the I / O scheduling device maintains the registration information of the application program and tracks the operation phase of the running application program.
[0141] S504, the I / O scheduling device maintains the status of the cache nodes and the available cache capacity included in the buffer device.
[0142] S505, the I / O scheduling device maintains scheduling queue information. The scheduling queue information includes scheduled but unexecuted I / O requests and unscheduled I / O requests. The I / O scheduling device updates the scheduling queue information according to the operation phase of the maintained application program.
[0143] S506: When the application is started, the access device registers the job information of the application with the I / O scheduling device.
[0144] S507, after receiving the registration information corresponding to the application, the I / O scheduling device calculates the interference value and sensitivity of the a posteriori indicator. The time for the I / O scheduling device to calculate the interference value and sensitivity is arbitrary. In S507, after receiving the registration information corresponding to the application, the I / O scheduling device calculates the interference value and sensitivity of the a posteriori indicator, which is only one way to calculate the a posteriori indicator.
[0145] S508, the I / O scheduling device generates a target scheduling policy corresponding to a plurality of I / O requests to be scheduled according to a dynamic programming of minimizing cost, wherein the plurality of I / O requests to be scheduled are I / O requests in a scheduling queue.
[0146] about Figure 5 For the implementation of each step in the corresponding embodiment, please refer to Figure 2 , Figure 3 and Figure 4 The corresponding embodiments will not be described in detail here.
[0147] The technical solution provided by the embodiment of the present disclosure utilizes dynamic programming that minimizes scheduling costs to generate a target scheduling strategy for scheduling multiple I / O requests to be scheduled. The target scheduling strategy generated using the data volume of the I / O requests to be scheduled and the available cache capacity of the reference cache node takes into account the available cache capacity representing the state of the reference cache node and the data volume representing the state of the I / O requests to be scheduled, so that the generated target scheduling strategy can better adapt to the dynamic changes of the buffer where the reference cache node is located.
[0148] In addition, the target scheduling strategy generated by dynamic programming based on minimizing the scheduling cost can improve the operating efficiency of the buffer system and the utilization of buffer resources, thereby reducing the probability of significant delays or interruptions in the operation of the application corresponding to the I / O request to be scheduled.
[0149] In addition, the technical solution provided in the embodiment of the present disclosure also proposes two a posteriori indicators (sensitivity and interference value) and calculation methods to quantitatively describe the degree of slowdown of the application due to interference, providing a reference for evaluating the performance of cache nodes and the effects of different scheduling strategies.
[0150] Based on the same inventive concept, the embodiment of the present disclosure also provides an I / O request scheduling device, as described in the following embodiment. Since the principle of solving the problem in the device embodiment is similar to that in the above method embodiment, the implementation of the device embodiment can refer to the implementation of the above method embodiment, and the repeated parts will not be repeated.
[0151] Figure 6 A schematic diagram of an I / O request scheduling device in an embodiment of the present disclosure is shown. Figure 6As shown, the device includes: an acquisition module 601, used to obtain multiple I / O requests to be scheduled and the corresponding data volume; the acquisition module 601 is also used to obtain the available cache capacity of multiple reference cache nodes; a generation module 602 is used to generate a target scheduling strategy corresponding to the data volume and the available cache capacity based on dynamic planning that minimizes the scheduling cost; a scheduling module 603 is used to schedule multiple I / O requests to be scheduled to multiple reference cache nodes according to the target scheduling strategy.
[0152] In one embodiment of the present disclosure, an acquisition module 601 is used to obtain job information of a first application, where the first application is an application that has completed registration with a scheduling device, and the job information includes a first I / O request of the first application and a corresponding data volume; determine a second application in the first application whose job phase is an I / O phase; determine a second I / O request in the first I / O request that causes the second application to be in the I / O phase; and determine multiple I / O requests to be scheduled and corresponding data volumes based on the second I / O request and the job information.
[0153] In one embodiment of the present disclosure, an acquisition module 601 is used to obtain available cache capacities corresponding to multiple cache nodes to be selected, where the multiple cache nodes to be selected are cache nodes included in a buffer device for receiving I / O requests; and determine that a cache node whose available cache capacity is greater than a capacity threshold among the multiple cache nodes to be selected is a reference cache node.
[0154] In one embodiment of the present disclosure, the acquisition module 601 is used to receive an indication message sent by a buffer device, where the indication message includes available cache capacities corresponding to multiple reference cache nodes, and the available cache capacities corresponding to the multiple reference cache nodes are all greater than a capacity threshold; based on the indication message, the available cache capacity of the reference cache node is determined.
[0155] In one embodiment of the present disclosure, a generation module 602 is used to determine the data volume and available cache capacity, and one or more corresponding intermediate scheduling costs under one or more dynamic programming methods for minimizing the scheduling cost; determine the minimum scheduling cost among the one or more intermediate scheduling costs; and determine the scheduling strategy corresponding to the minimum scheduling cost as the target scheduling strategy.
[0156] In one embodiment of the present disclosure, the number of the plurality of I / O requests to be scheduled is n, the number of the plurality of reference cache nodes is m, and both n and m are integers greater than or equal to 2; and the one or more dynamic programming methods for minimizing the scheduling cost include one or more of the following dynamic programming methods:
[0157] C ij +OPT(n -i ,m -j )
[0158] OPT -i ,m)
[0159] OPT(n,m -j )
[0160] OPT -i ,m -j )
[0161] Among them, C ij is the scheduling cost of scheduling the i-th I / O request to be scheduled to the j-th reference cache node; OPT(n -i ,m -j ) is the minimum scheduling cost of scheduling the other I / O requests to be scheduled except the i-th I / O request to be scheduled among the n I / O requests to be scheduled to the other cache nodes except the j-th reference cache node among the m reference cache nodes based on dynamic programming; OPT(n -i OPT(n,m) is the minimum scheduling cost of scheduling the n I / O requests to be scheduled except the i-th I / O request to be scheduled to m reference cache nodes based on dynamic programming; OPT(n,m -j ) is based on dynamic programming, and is a method for scheduling n I / O requests to be scheduled to other reference cache nodes among m reference cache nodes except the j-th reference cache node.
[0162] In one embodiment of the present disclosure, when the target scheduling policy corresponds to maximizing the bandwidth obtained by the application in the buffer device, C ij The completion time is determined based on the number of multiple I / O requests to be scheduled, the corresponding data volume, and the completion time; the completion time is the time taken by multiple reference cache nodes to complete the corresponding I / O requests to be scheduled.
[0163] In one embodiment of the present disclosure, when the target scheduling strategy corresponds to the maximum buffer system efficiency, C ij according to and At least one of the following is determined; is the average value of the sum of the ratios of each service duration to the previous service duration in the multiple service durations, the multiple service durations being the durations during which the multiple reference cache nodes provide services for the multiple third applications; the multiple third applications are the applications corresponding to the multiple I / O requests to be scheduled; is the average remaining cache capacity of multiple reference cache nodes.
[0164] In one embodiment of the present disclosure, when the objective of the target scheduling strategy is to maximize the fairness of the application program, C ijDetermined based on the first completion time and the second completion time; the first completion time is the time when multiple reference cache nodes complete services for the first third application; the second completion time is the time when multiple reference cache nodes complete services for the last third application; wherein, the first third application is the first third application registered with the scheduling device among multiple third applications; the last third application is the last third application registered with the scheduling device among multiple third applications.
[0165] In one embodiment of the present disclosure, when the purpose of the target scheduling strategy is to minimize the delay duration of the application program, C ij The waiting scheduling time and the additional time corresponding to each I / O request to be scheduled are determined according to multiple I / O requests to be scheduled; the waiting scheduling time is the difference between the time point when multiple reference cache nodes serve the corresponding I / O request to be scheduled and the time point when they serve the first I / O request to be scheduled among the multiple I / O requests to be scheduled; the additional time is the difference between the time when multiple reference cache nodes serve the corresponding I / O request to be scheduled and the time when multiple reference cache nodes serve the corresponding I / O request to be scheduled in a non-interference situation.
[0166] In one embodiment of the present disclosure, the acquisition module 601 is also used to obtain the concurrency tolerance of the cache node to be detected, which indicates the concurrency capability of the cache node to be detected; obtain the concurrent number of applications concurrently accessed in the cache node to be detected; obtain the achievable maximum bandwidth and actual bandwidth of the cache node to be detected; the device also includes: a first determination module 604, which is used to determine the sensitivity of the cache node to be detected to concurrent access based on the concurrency tolerance, the concurrent number, the achievable maximum bandwidth and the actual bandwidth.
[0167] In one embodiment of the present disclosure, the concurrency tolerance includes one or more of the bus channels, controller channels, and the number of storage bodies corresponding to the cache node to be detected.
[0168] In one embodiment of the present disclosure, the first determination module 604 is configured to calculate the sensitivity according to the following formula:
[0169]
[0170]
[0171] δ=λ τ ×λ ω
[0172] Among them, τ a is the number of concurrent connections; τ dc is the concurrency tolerance; ω max To achieve the maximum bandwidth;r is the actual bandwidth; δ is the sensitivity.
[0173] In one embodiment of the present disclosure, the acquisition module 601 is also used to obtain the first execution duration of the job corresponding to the application to be detected under interference conditions, where interference means that other applications are executed in parallel in the buffer device; and to obtain the second execution duration of the job under no interference conditions, where no interference means that no other applications are executed in parallel in the buffer device; the device also includes: a second determination module 605, which is used to determine the interference value based on the first execution duration and the second execution duration.
[0174] In one embodiment of the present disclosure, the second determination module 605 is configured to calculate the interference value according to the following formula:
[0175]
[0176] Among them, ET with_interference is the first execution time; ET without_interference is the second execution time; ε is the interference value.
[0177] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as "circuits", "modules" or "systems".
[0178] Refer to the following Figure 7 An electronic device 700 according to this embodiment of the present disclosure is described. Figure 7 The electronic device 700 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.
[0179] like Figure 7 As shown, the electronic device 700 is in the form of a general computing device. The components of the electronic device 700 may include but are not limited to: at least one processing unit 710, at least one storage unit 720, and a bus 730 connecting different system components (including the storage unit 720 and the processing unit 710).
[0180] The storage unit stores program codes, which can be executed by the processing unit 710, so that the processing unit 710 executes the steps according to various exemplary embodiments of the present disclosure described in the above “Specific Implementation Methods” section of this specification.
[0181] The storage unit 720 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 7201 and / or a cache storage unit 7202 , and may further include a read-only storage unit (ROM) 7203 .
[0182] The storage unit 720 may also include a program / utility 7204 having a set (at least one) of program modules 7205, such program modules 7205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.
[0183] Bus 730 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.
[0184] The electronic device 700 may also communicate with one or more external devices 740 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable a user to interact with the electronic device 700, and / or any device that enables the electronic device 700 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed through an input / output (I / O) interface 750. Furthermore, the electronic device 700 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 760. Figure 7 As shown, the network adapter 760 communicates with other modules of the electronic device 700 via the bus 730. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.
[0185] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.
[0186] In an exemplary embodiment of the present disclosure, a computer-readable storage medium is also provided, which may be a readable signal medium or a readable storage medium. A program product capable of implementing the above-mentioned method of the present disclosure is stored thereon. In some possible implementations, various aspects of the present disclosure may also be implemented in the form of a program product, which includes a program code, and when the program product is run on a terminal device, the program code is used to cause the terminal device to execute the steps according to various exemplary implementations of the present disclosure described in the above “Specific Implementation Methods” section of this specification.
[0187] More specific examples of computer-readable storage media in the present disclosure may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0188] In the present disclosure, a computer readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, wherein a readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A readable signal medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0189] Alternatively, the program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.
[0190] In a specific implementation, the program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, etc., and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).
[0191] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.
[0192] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.
[0193] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the implementation of the present disclosure.
[0194] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The description and examples are to be regarded as exemplary only, and the true scope of the present disclosure is indicated by the appended claims.
Claims
1. A method for scheduling input / output (I / O) requests, characterized in that: include: Get multiple I / O requests to be scheduled and the corresponding data volume; Obtain available cache capacities of multiple reference cache nodes; Generate a target scheduling strategy corresponding to the data volume and the available cache capacity based on dynamic programming that minimizes scheduling costs; Scheduling the multiple I / O requests to be scheduled to the multiple reference cache nodes according to the target scheduling policy; Among them, the dynamic programming based on minimizing scheduling costs generates the target scheduling strategy corresponding to the data volume and the available cache capacity, including: determining one or more intermediate scheduling costs corresponding to the data volume and the available cache capacity under one or more dynamic programming methods for minimizing scheduling costs; determining the minimum scheduling cost among the one or more intermediate scheduling costs; and determining the scheduling strategy corresponding to the minimum scheduling cost as the target scheduling strategy.
2. The method according to claim 1, characterized in that The obtaining of multiple I / O requests to be scheduled and corresponding data volumes includes: Acquire job information of a first application, where the first application is an application that has completed registration with a scheduling device, and the job information includes a first I / O request of the first application and a corresponding data volume; Determine a second application program in the first application program whose operation phase is an I / O phase; Determine a second I / O request in the first I / O request that causes the second application to be in the I / O phase; According to the second I / O request and the job information, a plurality of I / O requests to be scheduled and corresponding data amounts are determined.
3. The method according to claim 1, characterized in that The obtaining of available cache capacities of multiple reference cache nodes includes: Obtaining available cache capacities corresponding to a plurality of cache nodes to be selected, wherein the plurality of cache nodes to be selected are cache nodes included in a buffer device for receiving I / O requests; A cache node whose available cache capacity is greater than a capacity threshold among the multiple cache nodes to be selected is determined as a reference cache node.
4. The method according to claim 1, characterized in that: The obtaining of available cache capacities of multiple reference cache nodes includes: receiving an indication message sent by a buffer device, the indication message including available cache capacities corresponding to the multiple reference cache nodes, the available cache capacities corresponding to the multiple reference cache nodes all being greater than a capacity threshold; According to the indication message, an available cache capacity of the reference cache node is determined.
5. The method according to claim 1, characterized in that The number of the multiple I / O requests to be scheduled is n, the number of the multiple reference cache nodes is m, and n and m are both integers greater than or equal to 2; the one or more dynamic programming methods for minimizing the scheduling cost include one or more of the following dynamic programming methods: C ij +OPT(n -i ,m -j ) OPT(n -i ,m) OPT(n,m -j ) OPT(n -i ,m -j ) Among them, C ij is the scheduling cost of scheduling the i-th I / O request to be scheduled to the j-th reference cache node; OPT(n -i ,m -j ) is a minimum scheduling cost for scheduling other I / O requests to be scheduled among n I / O requests to be scheduled except the i-th I / O request to be scheduled to other cache nodes among m reference cache nodes except the j-th reference cache node based on dynamic programming; OPT -i ,m) is the minimum scheduling cost of scheduling the other I / O requests to be scheduled except the i-th I / O request to be scheduled among the n I / O requests to be scheduled to m reference cache nodes based on dynamic programming; OPT(n,m -j ) is based on dynamic programming, and is a method for scheduling n I / O requests to be scheduled to other reference cache nodes among m reference cache nodes except the j-th reference cache node.
6. The method according to claim 5, characterized in that When the target scheduling strategy corresponds to maximizing the bandwidth obtained by the application in the buffer device, the C ij Determining according to the number of the plurality of I / O requests to be scheduled and the corresponding data volumes and completion times; The completion duration is the duration used by the multiple reference cache nodes to complete the corresponding I / O requests to be scheduled.
7. The method according to claim 5, characterized in that When the target scheduling strategy corresponds to the maximum buffer system efficiency, the C ij according to and At least one of the following is determined; Said is the average value of the sum of the ratios of each service duration to the previous service duration in a plurality of service durations, wherein the plurality of service durations are the durations during which the plurality of reference cache nodes provide services for a plurality of third applications; the plurality of third applications are applications corresponding to the plurality of I / O requests to be scheduled; is the average remaining cache capacity of the multiple reference cache nodes.
8. The method according to claim 5, characterized in that When the objective of the target scheduling strategy is to maximize the fairness of the application, the C ij Determined based on the first completion time and the second completion time; The first completion time is the time when the multiple reference cache nodes complete the service for the first third application; the second completion time is the time when the multiple reference cache nodes complete the service for the last third application; The first third application is the first third application registered with the scheduling device among multiple third applications; and the last third application is the last third application registered with the scheduling device among the multiple third applications.
9. The method according to claim 5, characterized in that When the target scheduling strategy corresponds to minimizing the application delay duration, the C ij Determine, according to the multiple I / O requests to be scheduled, the waiting scheduling time and the additional time corresponding to each I / O request to be scheduled; The waiting scheduling duration is the difference between the time point at which the multiple reference cache nodes serve the corresponding I / O requests to be scheduled and the time point at which the multiple reference cache nodes serve the first I / O request to be scheduled among the multiple I / O requests to be scheduled; The additional time duration is the difference between the time duration that the multiple reference cache nodes serve the corresponding I / O requests to be scheduled and the time duration that the multiple reference cache nodes serve the corresponding I / O requests to be scheduled in the absence of interference.
10. The method according to any one of claims 1 to 9, characterized in that: Also includes: Acquire the concurrency tolerance of the cache node to be detected, where the concurrency tolerance indicates the concurrency capability of the cache node to be detected; Obtaining the concurrent number of applications concurrently accessed in the cache node to be detected; Obtaining the achievable maximum bandwidth and actual bandwidth of the cache node to be detected; The sensitivity of the cache node to be detected to concurrent access is determined according to the concurrency tolerance, the concurrency quantity, the maximum achievable bandwidth and the actual bandwidth.
11. The method according to claim 10, characterized in that The concurrency tolerance includes one or more of the bus channels, controller channels, and the number of storage bodies corresponding to the cache node to be detected.
12. The method according to claim 10, characterized in that The determining, according to the concurrency tolerance, the concurrency quantity, the achievable maximum bandwidth and the actual bandwidth, the sensitivity of the cache node to be detected to concurrent access includes: The sensitivity is calculated according to the following formula: d=λ τ ×λ ω Among them, τ a is the number of concurrent connections; τ dc is the concurrency tolerance; ω max is the maximum achievable bandwidth; ω r is the actual bandwidth; δ is the sensitivity.
13. The method according to any one of claims 1 to 9, characterized in that: Also includes: Obtaining a first execution duration of a job corresponding to the application to be detected under interference, wherein interference indicates that other applications are executed in parallel in the buffer device; Obtaining a second execution duration of the job in a non-interference situation, where the non-interference situation means that there are no other applications executed in parallel in the buffer device; An interference value is determined according to the first execution duration and the second execution duration.
14. The method according to claim 13, characterized in that The determining the interference value according to the first execution duration and the second execution duration includes: The interference value is calculated according to the following formula: Among them, E.T. with_interference The first execution time; ET without_interference is the second execution duration; and ε is the interference value.
15. An input / output (I / O) request scheduling device, characterized in that: include: An acquisition module is used to acquire multiple I / O requests to be scheduled and the corresponding data volume; The acquisition module is further used to obtain the available cache capacity of multiple reference cache nodes; A generation module, configured to generate a target scheduling strategy corresponding to the data volume and the available cache capacity based on dynamic programming that minimizes scheduling costs; A scheduling device, configured to schedule the multiple I / O requests to be scheduled to the multiple reference cache nodes according to the target scheduling policy; The generating module is further used to determine one or more intermediate scheduling costs corresponding to the data volume and the available cache capacity under one or more dynamic programming methods for minimizing scheduling costs; Determine a minimum scheduling cost among the one or more intermediate scheduling costs; and determine a scheduling strategy corresponding to the minimum scheduling cost as the target scheduling strategy.
16. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to execute the input / output (I / O) request scheduling method according to any one of claims 1 to 14 by executing the executable instructions.
17. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the input / output (I / O) request scheduling method according to any one of claims 1 to 14 is implemented.
Citation Information
Patent Citations
Random I / O scheduling method based on computing platform cache state perception
CN113127196A
Adaptive micro-service scheduling system for data I / O optimization
CN114281512A
ETL Scheduling Method and Apparatus Based on Deep Learning Algorithm
CN114936085A
Fast alignment of large-scale sequences using linear space techniques
US20070076936A1
