CN118152125A - Computing power resource scheduling system based on cloud platform - Google Patents
Computing power resource scheduling system based on cloud platform Download PDFInfo
- Publication number
- CN118152125A CN118152125A CN202410302642.0A CN202410302642A CN118152125A CN 118152125 A CN118152125 A CN 118152125A CN 202410302642 A CN202410302642 A CN 202410302642A CN 118152125 A CN118152125 A CN 118152125A
- Authority
- CN
- China
- Prior art keywords
- service
- computing power
- power
- unpacking
- demand
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004364 calculation method Methods 0.000 claims abstract description 192
- 238000012545 processing Methods 0.000 claims abstract description 38
- 238000013507 mapping Methods 0.000 claims abstract description 18
- 238000003032 molecular docking Methods 0.000 claims abstract description 12
- 238000004806 packaging method and process Methods 0.000 claims abstract description 8
- 230000000875 corresponding effect Effects 0.000 claims description 93
- 230000002596 correlated effect Effects 0.000 claims description 4
- 230000001276 controlling effect Effects 0.000 claims description 3
- 238000000034 method Methods 0.000 description 11
- 238000012856 packing Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention relates to the technical field of computer power dispatching processing, in particular to a cloud platform-based power resource dispatching system, which comprises a service demand docking station, a power dispatching system and a power dispatching system, wherein the service demand docking station acquires and determines the power demand corresponding to the power service demand; the computing power cloud platform is provided with a plurality of computing power nodes so as to monitor the current computing power allowance of each computing power node; the service unpacking station is used for analyzing the single calculation service requirement into a plurality of parallel calculation service packages according to the unpacking judgment result and the service logic; the service packaging station packages and outputs each service completion packet corresponding to the single computing power service requirement according to the mapping relation; the computing power cloud scheduling station is used for determining the number of enabled corresponding computing power nodes according to the computing power business requirements; the invention effectively reduces the single-packet calculation force demand of the parallel calculation force service packet received by the single calculation force node, and improves the calculation processing speed of the single calculation force node, thereby improving the processing speed of the overall calculation force service demand of the user.
Description
Technical Field
The invention relates to the technical field of computer power scheduling, in particular to a power resource scheduling system based on a cloud platform.
Background
Data, computational effort and algorithms are three elements of artificial intelligence development. The calculation power refers to the calculation capability of realizing the output of a target result by processing information data. The calculation force scheduling is to balance the demand and supply of calculation force in a certain range through the scheduling of calculation force. Cloud computing is a new attempt of distributed computing, and the essence of the cloud computing is to package and aggregate a large amount of scattered computing power resources, so that computing power with higher reliability, higher performance and lower cost is realized. Therefore, the computing power resources can be integrated in a larger range, and higher computing power resource allocation is realized.
Chinese patent publication No. CN115460216a discloses a method and apparatus for scheduling computational power resources, a device and system for scheduling computational power resources; relates to the technical field of computers. When the node types of the candidate computing nodes are the same, acquiring the resource use condition and the energy use condition of the candidate computing nodes; determining the effective load of each candidate computing node according to the resource use condition and the energy use condition of the candidate computing node; acquiring the total amount of calculation force demand corresponding to the user service demand, and distributing bearing service volume for each candidate calculation force node according to the total amount of calculation force demand and the effective load of each candidate calculation force node; and scheduling the corresponding candidate computing nodes to provide corresponding computing resources for the user according to the bearing traffic of each candidate computing node. Therefore, the technical scheme can not realize the dispatching of the calculation force under the condition that the calculation force service requirement exceeds the maximum effective load of a single calculation force node, so that the calculation force waste is caused.
Disclosure of Invention
Therefore, the invention provides a computing power resource scheduling system based on a cloud platform, which is used for solving the problem that computing power scheduling cannot be performed under the condition that the single-packet computing power service requirement exceeds the maximum effective load of a single computing power node in the prior art.
In order to achieve the above object, the present invention provides a computing power resource scheduling system based on a cloud platform, including:
the business demand docking station is used for acquiring the power calculation business demands of the users and determining the power calculation demand corresponding to each power calculation business demand;
the computing power cloud platform is provided with a plurality of computing power nodes and is connected with the business demand docking station to monitor the current computing power allowance of each computing power node;
The service unpacking station is respectively connected with the service demand docking station and the computing power cloud platform and is used for calculating the prediction processing time length according to the computing power demand, determining whether the single service demand needs unpacking according to the current computing power allowance and the prediction processing time length of each computing power node, analyzing the single computing power service demand into a plurality of parallel computing power service packages according to service logic corresponding to the computing power service demand according to unpacking judging results, and setting the mapping relation among the parallel computing power service packages;
the service packaging station is respectively connected with the service unpacking station and the power cloud platform and is used for receiving service completion packets completed by all power computing nodes, matching all service completion packets corresponding to single power computing service requirements according to the service logic and packaging and outputting all service completion packets corresponding to the single power computing service requirements according to the mapping relation;
The computing power cloud dispatching station is connected with the computing power cloud platform and used for determining the number of enabled corresponding computing power nodes according to computing power service requirements, and controlling the computing power cloud platform to match each service packet to the corresponding computing power node according to the number of packets of a single service requirement and the single packet computing power requirement;
the computing power cloud platform at least comprises three computing power nodes corresponding to the computing power business requirements.
Further, the computing power cloud dispatching station determines the quantity adjustment mode of the currently-started computing power nodes according to the corresponding relation between the computing power demand total quantity corresponding to the current computing power service demand and the current computing power total allowance, wherein,
And if the total amount of the computing power demands corresponding to the current computing power business demands is more than or equal to 0.5 times of the total amount of the current computing power, the computing power cloud dispatching station judges that the starting number of the computing power nodes is increased.
Further, the service unpacking station is preset with unpacking rules, and includes:
determining whether the single service demand needs unpacking or not according to the ratio of the maximum value in the current computing power allowance of each enabled computing power node to the computing power demand corresponding to the single service demand;
and determining whether the single service requirement needs to be unpacked or not according to the predicted processing time length.
Further, a plurality of preset unpacking logic points corresponding to the service logic are preset in the service unpacking station, the service unpacking station analyzes the single calculation service requirement into the service logic, and a plurality of unpacking selectable points are determined according to the corresponding relation between the service logic and each preset unpacking logic point.
Further, the calculation force logic corresponding to the business logic before and after the preset unpacking logic point can be independently operated, and the calculation force demand of the preset unpacking logic point is lower than the preset calculation force demand.
Further, the service unpacking station determines the expected unpacking quantity according to the ratio of the maximum unpacking quantity formed by the unpacking optional points of the single calculation power service requirement to the calculation power requirement quantity, and determines the unpacking point according to the expected unpacking quantity.
Further, the service unpacking station analyzes the single power service requirement into a plurality of parallel power service packages according to the unpacking point determined by the single power service requirement, and configures the mapping relation among the parallel power service packages according to the sequence of the unpacking points on the service logic;
the mapping relation is the operation relation between the calculation force end point of the former parallel calculation force service packet and the calculation force start point of the adjacent latter parallel calculation force service packet in service logic.
Further, the computing power cloud scheduling station determines matched corresponding computing power nodes according to the computing power demand of each single packet of single service demand;
Wherein the single-packet computing force demand is positively correlated with the current computing force margin of the matched corresponding computing force node.
Further, the service unpacking station determines an adjustment mode of the unpacking quantity according to a comparison result of the total calculated force demand in the current preset time period and the average calculated force demand in the corresponding historical time period.
Further, the computing power cloud scheduling station is further configured to determine a number adjustment mode of the computing power nodes according to a current computing power total allowance of each computing power node and a historical computing power demand total amount corresponding to a current period.
Compared with the prior art, the method has the beneficial effects that the unpacking judgment is carried out on the calculation power service demands of the users, so that the single-package calculation power demand of the parallel calculation power service packages received by the single calculation power node is effectively reduced, the calculation processing speed of the single calculation power node is improved, and the processing speed of the overall calculation power service demands of the users is improved.
Furthermore, the unpacking points are determined through the calculation force demand, the calculation force demand of the preset unpacking logic points is lower than the preset calculation force demand, calculation force occupation in the unpacking process of the finished packets of each service is effectively avoided, the packing efficiency is improved, the mapping relation among the parallel calculation force service packets is configured through the sequence of the unpacking points on the service logic, the effectiveness and the integrity of operation in the unpacking points are effectively ensured, the calculation force logic corresponding to the service logic before and after the preset unpacking logic points can be independently operated, the situation that strong association operation is set as the unpacking point is avoided, and the effectiveness and the integrity of operation in the unpacking points are further improved.
Further, the invention determines the adjustment mode of the unpacking quantity according to the comparison result of the total calculation force demand in the current preset time period and the average calculation force demand in the corresponding historical time period, and determines the quantity adjustment mode of the calculation force nodes according to the total current calculation force allowance of each calculation force node and the total historical calculation force demand corresponding to the current time period, so that the calculation force can be better adapted to the current calculation force demand, and the calculation force is scheduled in advance in combination with the historical calculation force demand trend, thereby further improving the calculation force scheduling efficiency of the calculation force cloud platform.
Drawings
FIG. 1 is a schematic diagram of a computing power resource scheduling system based on a cloud platform according to an embodiment of the present invention;
FIG. 2 is a logic diagram of unpacking a single power service requirement according to an embodiment of the present invention;
FIG. 3 is a schematic diagram showing the relationship between service logic and unpacking logic points according to an embodiment of the present invention;
Fig. 4 is an unpacking schematic diagram of a parallel computing power service packet according to an embodiment of the present invention.
Detailed Description
In order that the objects and advantages of the invention will become more apparent, the invention will be further described with reference to the following examples; it should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are merely for explaining the technical principles of the present invention, and are not intended to limit the scope of the present invention.
It should be noted that, in the description of the present invention, terms such as "upper," "lower," "left," "right," "inner," "outer," and the like indicate directions or positional relationships based on the directions or positional relationships shown in the drawings, which are merely for convenience of description, and do not indicate or imply that the apparatus or elements must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present invention.
Furthermore, it should be noted that, in the description of the present invention, unless explicitly specified and limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly, and may be either fixedly connected, detachably connected, or integrally connected, for example; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention can be understood by those skilled in the art according to the specific circumstances.
For convenience of description, terms appearing in the embodiments are explained as follows:
calculating force refers to calculating capability, which refers to processing capability of data, and the magnitude of the calculating force represents the strength of the processing capability of the digital information data;
And (3) calculating force scheduling: the calculation force is scheduled, so that the demand and the supply of the calculation force reach balance within a certain range, and the calculation force scheduling method comprises two calculation force scheduling modes of scheduling calculation tasks among calculation force equipment, configuring and adjusting the thread number of the equipment (machine tool) and the state of a model GPU/CPU, and reducing the time consumption of an equipment link;
calculating force demand: the method is the calculation task amount processed by a computer;
business logic: an operation logic for analyzing according to the operation force service requirement;
Unpacking: decomposing the overall power calculation business requirement into a plurality of independent power calculation business requirements;
Packaging: and calculating the calculation result of each calculation power service requirement after unpacking according to the mapping relation, and outputting the calculation result of the whole calculation power service requirement.
Referring to fig. 1, which is a schematic structural diagram of a computing power resource scheduling system based on a cloud platform according to an embodiment of the present invention, the embodiment of the present invention provides a computing power resource scheduling system based on a cloud platform, including:
the business demand docking station is used for acquiring the power calculation business demands of the users and determining the power calculation demand corresponding to each power calculation business demand;
the computing power cloud platform is provided with a plurality of computing power nodes and is connected with the business demand docking station to monitor the current computing power allowance of each computing power node;
The business unpacking station is respectively connected with the business demand docking station and the computing power cloud platform and is used for calculating the prediction processing time length according to the computing power demand, determining whether the single business demand needs unpacking according to the current computing power allowance and the prediction processing time length of each computing power node, analyzing the single computing power business demand into a plurality of parallel computing power business packages according to unpacking judgment results and business logic, and setting the mapping relation among the parallel computing power business packages;
the service packaging station is respectively connected with the service unpacking station and the power cloud platform and is used for receiving service completion packets completed by all power computing nodes, matching all service completion packets corresponding to single power computing service requirements according to the service logic and packaging and outputting all service completion packets corresponding to the single power computing service requirements according to the mapping relation;
The computing power cloud dispatching station is connected with the computing power cloud platform and used for determining the number of enabled corresponding computing power nodes according to computing power service requirements, and controlling the computing power cloud platform to match each service packet to the corresponding computing power node according to the number of packets of a single service requirement and the single packet computing power requirement;
the computing power cloud platform at least comprises three computing power nodes corresponding to the computing power business requirements.
The invention effectively reduces the single-packet calculation force demand of the parallel calculation force service packet received by a single calculation force node by unpacking and judging the calculation force service demand of the user, and improves the calculation processing speed of the single calculation force node, thereby improving the processing speed of the whole calculation force service demand of the user.
It is understood that the service requirement docking station includes, but is not limited to, an intelligent terminal, an edge computing device (such as an edge computing gateway), a cloud computing device, and the like, where the intelligent terminal may be any terminal computing resource scheduling device capable of providing computing resources, such as a mobile phone, a computer, an intelligent voice interaction device, an intelligent home appliance, a vehicle-mounted terminal, a test terminal, a computer, and the like, and is not limited herein. The service unpacking station, the service packing station and the computing power cloud dispatching station can be independent physical servers respectively, and can also be a server cluster formed by a plurality of physical servers or a part of a distributed control system of the cloud platform, wherein each computing power node in the cloud platform, the service unpacking station, the service packing station and the computing power cloud dispatching station form a block chain, and support basic cloud computing services such as big data of the cloud platform, an artificial intelligent platform and the like.
Specifically, the computing power cloud dispatching station determines the number of corresponding computing power nodes according to the corresponding relation between the computing power demand total amount corresponding to the current computing power business demand and the current computing power total allowance, wherein,
If the total amount of the computing power demands corresponding to the current computing power business demands is more than or equal to 0.5 times of the total amount of the current computing power allowance, the computing power cloud scheduling station judges to increase the starting number of the corresponding computing power nodes;
Wherein, the increase of the starting number of the computing power nodes is positively correlated with the total computing power demand corresponding to the current computing power service demand.
In the implementation, the total current computing force margin is determined according to the sum of the current computing force margins of all the currently-started computing force nodes in the computing force cloud platform, and the computing force margin of all the computing force nodes can be calculated according to any mode of computing the computing force margin in the prior art, and the method is not limited herein.
According to the method, the corresponding relation between the total amount of the computing force demands corresponding to the current computing force service demands and the total allowance of the current computing force is judged to match the computing force processing capacity of the computing force cloud platform, the integral computing force allowance of the computing force cloud platform is insufficient, the starting quantity of the computing force nodes is determined through the computing force comparison, the problems that after the current computing force service demands occupy computing force, the residual computing force allowance is insufficient to support the fact that the timeliness of newly added computing force nodes is low and the processing efficiency of the computing force demands is low due to the fact that the follow-up computing force service demands are not enough are effectively avoided, and the computing force scheduling efficiency of the computing force cloud platform is improved.
Referring to fig. 2, a service unpacking station presets a unpacking rule, including:
the unpacking rule I is used for determining whether the single service requirement needs unpacking or not according to the ratio of the maximum value in the current computing force allowance corresponding to each enabled computing force node to the computing force requirement corresponding to the single service requirement;
and a second unpacking rule is used for determining whether the single service requirement needs unpacking according to the predicted processing time length.
It can be understood that the single computing node corresponds to the current computing margin of the self computing node, and in each enabled computing node, the maximum value in each current computing margin is compared with the computing force demand corresponding to the single service demand, and whether the single service demand needs unpacking or not is determined according to the ratio;
The ratio B is calculated by the following steps: ，/> For the maximum value in the respective current calculation force margin,/> The amount of power demand corresponding to a single business demand.
In implementation, the first unpacking rule or the second unpacking rule can be used independently to judge whether the single service requirement needs unpacking, or the first unpacking rule and the second unpacking rule can be used simultaneously to judge whether the single service requirement needs unpacking. The unpacking judgment result comprises the steps of judging that the single service requirement does not need unpacking and judging that the single service requirement needs unpacking.
In a specific embodiment, the unpacking rule one and the unpacking rule two are used for judging whether the single service requirement needs unpacking, a preset ratio B0 and a preset predicted processing time t0 are set, the calculated ratio is B, the predicted processing time is t, and the judging mode of the unpacking rule comprises:
If B is more than or equal to B0 and t is less than t0, judging that the single service requirement does not need unpacking;
if B is less than B0 or t is more than or equal to t0, the single service requirement is judged to need unpacking.
In a specific embodiment, only a single service requirement pair of the unpacking rule is used to determine whether unpacking is needed, and the unpacking rule is determined in a manner that includes:
if B is more than or equal to B0, judging that the single service requirement does not need unpacking;
If B is less than B0, the single service requirement is judged to need unpacking.
In a specific embodiment, only the second unpacking rule is used to determine whether the single service requirement needs unpacking, and the unpacking rule is determined by the following steps:
if t is less than t0, judging that the single service requirement does not need unpacking;
If t is more than or equal to t0, the single service requirement is judged to need unpacking.
It can be understood that the preset ratio B0 is determined according to the amount of calculation required to be reserved in the service scenario, in general, the value range of the preset ratio B0 is [1.5,3], and the value of the preferred preset ratio B0 is 1.8, so that the calculation node has a better calculation margin, and the processing performance of the calculation node can be fully exerted. The preset prediction processing time t0 can be set according to the proposed processing time of the power calculation service demand or the set standard power calculation service processing time, and can be set according to specific service scenes as a unpacking judgment condition.
In the implementation, the prediction processing duration is determined according to the computing power demand quantity corresponding to the computing power node processing capability and the single service demand, and preferably, the computing power node processing capability in the prediction processing duration is determined by the processing capability of the computing power node corresponding to the maximum value in each current computing power allowance.
The processing efficiency of the computing power business demands in the computing power nodes of the cloud platform can be effectively judged through the judgment of the unpacking mode, so that the processing condition judgment of the computing power business demands can be effectively carried out through unpacking rules, the single computing power business demands are decomposed into a plurality of parallel computing power business packages through unpacking and are processed through different computing power nodes, and the processing efficiency of the computing power business demands is improved through the allocation and the use of the computing power nodes.
Fig. 3 is a schematic diagram showing a relationship between service logic and unpacking logic points in an embodiment of the present invention, wherein a plurality of preset unpacking logic points corresponding to the service logic are preset in the service unpacking station, the service unpacking station analyzes a single calculation service requirement into the service logic, and determines a plurality of unpacking selectable points according to the corresponding relationship between the service logic and each preset unpacking logic point.
Specifically, the calculation force logic corresponding to the business logic before and after the preset unpacking logic point can be independently operated, and the calculation force demand of the preset unpacking logic point is lower than the preset calculation force demand.
Preferably, the pre-imputation force demand is set in accordance with the actual calculation of the traffic unpacking station. In implementation, determining an unpacking optional point according to whether a preset unpacking logic point exists in the service logic, and if the preset unpacking logic point exists in the service logic, determining the point as a corresponding unpacking optional point.
In fig. 3, the unpacking selectable points corresponding to the preset unpacking logic point include a logic point 1, a logic point 2 and a logic point 3, and the three logic points intercept the whole service logic corresponding to the single computing power service requirement into a service logic 1, a service logic 2, a service logic 3, a service logic 4 and a service logic 5.
In implementation, the service logic refers to actual operation logic corresponding to the service demand of the calculation force, the preset unpacking logic point is generally set to be simple operation, including simple mathematical operation (addition, subtraction, multiplication, division, combination and the like) and simple logic operation (sum, or non-judgment and the like), and the calculation force logic corresponding to the service logic before and after the preset unpacking logic point is set to be independently operated, so that different calculation force nodes can independently process a single parallel calculation force service packet, and the service packing station can be accelerated to package and output each service completion packet.
Specifically, the service unpacking station determines the expected unpacking quantity according to the ratio of the maximum unpacking quantity formed by the unpacking optional points of the single calculation power service requirement to the calculation power requirement quantity, and determines the unpacking point according to the expected unpacking quantity.
It will be appreciated that the unpacking selectable points of a single power service demand form separate service logic segments at both ends, and that the maximum amount refers to the maximum value of the power demands of each formed service logic segment. The expected unpacking number is determined by calculating the ratio of the maximum calculation amount to the total calculation force demand amount corresponding to the calculation force business demand of the user, the expected unpacking number=the total calculation force demand amount/the maximum calculation amount, and the expected unpacking number is an integer rounded up.
Specifically, the unpacking point is determined according to the unpacking quantity, in the implementation, the calculation force demand of each independent service logic section has the condition of different magnitudes, after the expected unpacking quantity is determined, the average calculation force demand of each logic section is calculated according to the whole calculation force demand and the expected unpacking quantity, and the calculation force demand of the combined service logic section does not exceed the average calculation force demand by combining the adjacent service logic sections with smaller calculation force demand, so that the determined service logic sections can determine the unpacking point and the unpacking quantity. As for the combination of adjacent service logic segments, it is preferable to use an equalization method to minimize the difference in the amount of calculation between the service logic segments.
Please refer to fig. 4, which is a schematic diagram of unpacking a parallel computing power service packet according to an embodiment of the present invention, wherein the service unpacking station analyzes a single computing power service requirement into a plurality of parallel computing power service packets according to the unpacking point determined according to the single computing power service requirement, and configures a mapping relationship between the parallel computing power service packets according to the sequence of the unpacking points on the service logic;
the mapping relation is the operation relation between the calculation force end point of the former parallel calculation force service packet and the calculation force start point of the adjacent latter parallel calculation force service packet in service logic.
According to the invention, the unpacked parallel computing power business package is calculated through the mapping relation, so that on one hand, the processing efficiency of the computing power business demand can be improved through unpacking advantages, and on the other hand, after each computing power node finishes the operation of a single parallel computing power business package, the output corresponding to the overall computing power business demand is calculated through the mapping relation, so that the calculation error caused by the error corresponding to the parallel computing power business package under the condition that a plurality of computing power business demands are processed simultaneously can be avoided, and the packing efficiency and the packing accuracy are improved.
Specifically, the computing power cloud scheduling station determines matched corresponding computing power nodes according to the computing power demand of each single packet of single service demand;
Wherein the single-packet computing force demand is positively correlated with the current computing force margin of the matched corresponding computing force node.
In implementation, each single-packet power demand of a single service demand refers to a power demand corresponding to each parallel power service packet formed by unpacking the single service demand. Firstly, arranging according to the current calculation margin of each enabled calculation node according to the order of magnitude, arranging the calculation force demand of each single package according to the order of magnitude, starting to match the calculation force demand maximum value in the parallel calculation force service package, matching the calculation force node corresponding to the current calculation force margin maximum value of each enabled calculation node for the calculation force demand maximum value in the parallel calculation force service package, matching the calculation force node corresponding to the current calculation force margin secondary maximum value of each enabled calculation node for the calculation force demand secondary maximum value in the parallel calculation force service package until the parallel calculation force service package is matched, or rearranging the current calculation force margin of each currently enabled calculation node according to the order of magnitude under the condition that the parallel calculation force service package is matched, and the parallel calculation force service package matched is not matched (at this moment, the matched parallel calculation force service package enters the processing, and the current calculation force margin needs to be updated) until the parallel calculation force service package and the calculation force node which need to be matched are repeated until the parallel calculation force service package which is matched is completed.
Specifically, the service unpacking station determines an adjustment mode of the unpacking quantity according to a comparison result of the total calculated force demand in the current preset time period and the average calculated force demand in the corresponding historical time period.
In implementation, the preset time period refers to a preset duration range up to the current moment, the total amount of calculation power demand in the current preset time period is calculated according to the statistical result of the calculation power demand corresponding to the historical calculation power service demand, and the adjustment mode of unpacking quantity is determined according to the comparison result of the total amount of calculation power demand and the average calculation power demand in the corresponding historical time period, and the method comprises the following steps:
the total calculation force demand is larger than the average calculation force demand in the corresponding historical time period, and the unpacking number is judged to be increased, so that the single calculation force operation time of a single calculation force node is reduced, and the single package operation efficiency is improved.
And the total calculation force demand is smaller than or equal to the average calculation force demand in the corresponding historical time period, and the unpacking quantity is judged to be unchanged, so that unpacking and packing efficiency is improved, and the calculation force scheduling efficiency of the calculation force cloud platform is improved.
Generally, the range of the preset time period is set to 0.5-4 hours, so that the influence condition of the historical calculation force demand on the subsequent calculation force adjustment can be reflected, the trend of the subsequent calculation force demand can be prejudged through the comparison of the average calculation force demand in the historical time period, and the processing mode of the current calculation force service demand can be accurately judged to adapt to the subsequent calculation force demand.
Specifically, the computing power cloud scheduling station is further configured to determine a number adjustment manner of computing power nodes according to a current computing power total allowance of each computing power node and a historical computing power demand total amount corresponding to a current period.
If the total allowance of the current calculation force is larger than or equal to the total amount of the historical calculation force demand corresponding to the current period, the calculation force cloud scheduling station judges that the number of calculation force nodes is not required to be adjusted;
If the total margin of the current computing power is smaller than the total demand of the historical computing power corresponding to the current period, the computing power cloud scheduling station judges that the number of computing power nodes needs to be increased;
The number of the calculated force nodes is inversely related to the total allowance of the current calculated force, and the calculated force cloud scheduling station recalculates the total allowance of the current calculated force after increasing the calculated force nodes so as to ensure the total amount of the historical calculated force corresponding to the current period of time when the total allowance of the current calculated force is more than 1.1 times.
It can be understood that the historical calculated force demand total amount can be determined according to the average value of the calculated force demand total amount of the historical period corresponding to the current period, where the corresponding relationship of the corresponding historical period can be corresponding in specific periods of days, weeks, months, years, and the like, which are not described herein.
According to the invention, the unpacking points are determined by the calculation force demand, the calculation force demand of the preset unpacking logic points is lower than the preset calculation force demand, so that calculation force occupation in the unpacking process of each service completion packet is effectively avoided, the packing efficiency is improved, the mapping relation among the parallel calculation force service packets is configured according to the sequence of each unpacking point on the service logic, the validity and the integrity of the operation in each unpacking point are effectively ensured, the calculation force logic corresponding to the service logic before and after the preset unpacking logic points can be independently operated, the strong correlation operation is prevented from being set as the unpacking point, and the validity and the integrity of the operation in each unpacking point are further improved.
And the adjustment mode of the unpacking quantity is determined according to the comparison result of the total calculation force demand in the current preset time period and the average calculation force demand in the corresponding historical time period, and the quantity adjustment mode of the calculation force nodes is determined according to the total current calculation force allowance of each calculation force node and the total historical calculation force demand corresponding to the current time period, so that the calculation force can be better adapted to the current calculation force demand, the calculation force is scheduled in advance in combination with the historical calculation force demand trend, and the calculation force scheduling efficiency of the calculation force cloud platform is further improved.
Thus far, the technical solution of the present invention has been described in connection with the preferred embodiments shown in the drawings, but it is easily understood by those skilled in the art that the scope of protection of the present invention is not limited to these specific embodiments. Equivalent modifications and substitutions for related technical features may be made by those skilled in the art without departing from the principles of the present invention, and such modifications and substitutions will be within the scope of the present invention.
Claims (10)
1. The utility model provides a power resource scheduling system based on cloud platform which characterized in that includes:
the business demand docking station is used for acquiring the power calculation business demands of the users and determining the power calculation demand corresponding to each power calculation business demand;
the computing power cloud platform is provided with a plurality of computing power nodes and is connected with the business demand docking station to monitor the current computing power allowance of each computing power node;
The service unpacking station is respectively connected with the service demand docking station and the computing power cloud platform and is used for calculating the prediction processing time length according to the computing power demand, determining whether the single service demand needs unpacking according to the current computing power allowance and the prediction processing time length of each computing power node, analyzing the single computing power service demand into a plurality of parallel computing power service packages according to service logic corresponding to the computing power service demand according to unpacking judging results, and setting the mapping relation among the parallel computing power service packages;
the service packaging station is respectively connected with the service unpacking station and the power cloud platform and is used for receiving service completion packets completed by all power computing nodes, matching all service completion packets corresponding to single power computing service requirements according to the service logic and packaging and outputting all service completion packets corresponding to the single power computing service requirements according to the mapping relation;
The computing power cloud dispatching station is connected with the computing power cloud platform and used for determining the number of enabled corresponding computing power nodes according to computing power service requirements, and controlling the computing power cloud platform to match each service packet to the corresponding computing power node according to the number of packets of a single service requirement and the single packet computing power requirement;
the computing power cloud platform at least comprises three computing power nodes corresponding to the computing power business requirements.
2. The system of claim 1, wherein the computing power cloud scheduler determines a number adjustment mode of currently enabled computing power nodes according to a correspondence between a total computing power demand corresponding to a current computing power business demand and a total computing power allowance, wherein,
And if the total amount of the computing power demands corresponding to the current computing power business demands is more than or equal to 0.5 times of the total amount of the current computing power, the computing power cloud dispatching station judges that the starting number of the computing power nodes is increased.
3. The computing power resource scheduling system of claim 1, wherein the service unpacking station is preset with unpacking rules, and the system comprises:
determining whether the single service demand needs unpacking or not according to the ratio of the maximum value in the current computing power allowance of each enabled computing power node to the computing power demand corresponding to the single service demand;
and determining whether the single service requirement needs to be unpacked or not according to the predicted processing time length.
4. A system according to claim 3, wherein a plurality of preset unpacking logic points corresponding to the service logic are preset in the service unpacking station, the service unpacking station analyzes the single power service requirement into the service logic, and determines a plurality of unpacking selectable points according to the corresponding relation between the service logic and each preset unpacking logic point.
5. The system according to claim 4, wherein the power logic corresponding to the business logic before and after the preset unpacking logic point is independently operable, and the power demand of the preset unpacking logic point is lower than the power demand of the preset power demand.
6. The computing power resource scheduling system of claim 4, wherein the service unpacking station determines an expected unpacking number based on a ratio of a maximum amount of unpacking formed at each unpacking selectable point of a single computing power service demand to a computing power demand thereof, and determines an unpacking point based on the expected unpacking number.
7. The system according to claim 6, wherein the service unpacking station parses a single power service requirement into a plurality of parallel power service packages according to the unpacking point determined by the single power service requirement, and configures a mapping relationship between the parallel power service packages according to a sequence of the unpacking points on the service logic;
the mapping relation is the operation relation between the calculation force end point of the former parallel calculation force service packet and the calculation force start point of the adjacent latter parallel calculation force service packet in service logic.
8. The computing power resource scheduling system of claim 2, wherein the computing power cloud scheduling station determines matched corresponding computing power nodes according to each single-packet computing power demand of a single service demand;
Wherein the single-packet computing force demand is positively correlated with the current computing force margin of the matched corresponding computing force node.
9. The system according to claim 6, wherein the service unpacking station determines the unpacking amount adjustment mode according to a comparison result of the total amount of the calculation power demand in the current preset time period and the average calculation power demand in the corresponding historical time period.
10. The computing power resource scheduling system of claim 1, wherein the computing power cloud scheduling station is further configured to determine a number adjustment manner of computing power nodes according to a current computing power total allowance of each computing power node and a historical computing power demand total amount corresponding to a current period.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410302642.0A CN118152125A (en) | 2024-03-18 | 2024-03-18 | Computing power resource scheduling system based on cloud platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410302642.0A CN118152125A (en) | 2024-03-18 | 2024-03-18 | Computing power resource scheduling system based on cloud platform |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118152125A true CN118152125A (en) | 2024-06-07 |
Family
ID=91296547
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410302642.0A Pending CN118152125A (en) | 2024-03-18 | 2024-03-18 | Computing power resource scheduling system based on cloud platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118152125A (en) |
-
2024
- 2024-03-18 CN CN202410302642.0A patent/CN118152125A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110099384A (en) | Resource regulating method is unloaded based on side-end collaboration more MEC tasks of multi-user | |
| CN110248417B (en) | Resource allocation method and system for communication service in power Internet of things | |
| US10901388B2 (en) | Method and system for creating energy demand model | |
| CN105075184A (en) | Method and system to represent the impact of load variation on service outage over multiple links | |
| CN114143270B (en) | Bandwidth adjustment method and device | |
| CN110113140B (en) | Calculation unloading method in fog calculation wireless network | |
| CN113115327B (en) | Method, device, equipment and storage medium for dynamically optimizing network performance | |
| CN111212108B (en) | Multi-user parallel migration method based on non-orthogonal multiple access and mobile edge computing | |
| CN110780986B (en) | Internet of things task scheduling method and system based on mobile edge computing | |
| US8341266B2 (en) | Method and system for load balancing over a set of communication channels | |
| CN116700920A (en) | Cloud primary hybrid deployment cluster resource scheduling method and device | |
| CN102404778B (en) | Load estimation method | |
| CN113676423A (en) | Port flow control method and device, exchange chip and storage medium | |
| CN113596106A (en) | Time delay sensitive type slice resource scheduling method based on multi-edge server cooperation | |
| CN118152125A (en) | Computing power resource scheduling system based on cloud platform | |
| CN110366194B (en) | Resource calling method, device and system | |
| CN112437449A (en) | Joint resource allocation method and area organizer | |
| US20060072453A1 (en) | Method and apparatus for assessing traffic load of a communication network | |
| CN115150892B (en) | VM-PM repair strategy method in MEC wireless system with bursty traffic | |
| CN107995312B (en) | Boundary network service flow distribution system and dynamic distribution method | |
| KR102492716B1 (en) | User Equipment for Dynamically Offloading Tasks Based on Multi-Radio Access Technology for IIot Service and Offloading Method Thereof | |
| CN112492635B (en) | Determination method and device for scheduling priority | |
| KR102389665B1 (en) | Apparatus and method of constructing energy efficient communication and computation resource allocation framework for multiple communication service | |
| CN110493873B (en) | Wireless private network spectrum allocation optimization method and device suitable for power service | |
| US5941948A (en) | Admission controller for network access |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination |