A virtual machine network resource allocation system

Abstract

The present invention discloses a virtual machine network resource allocation system, which obtains habit data of a target object through a single acquisition unit, calculates a required computing power range of the target object according to the habit data, and obtains the priority of the target object by performing demand analysis on the habit data of the target object; a resource time control unit monitors the real-time computing power of the virtual machine and communicates with a processor; an abnormality identification unit performs abnormality analysis on non-target objects according to the habit data of the target object, and divides the non-target objects into abnormal target objects and normal target objects; the processor allocates resources to normal target objects according to the priority, the required computing power range and the real-time computing power; allocates network resource priority to each target object to ensure the throughput of the target object and the service quality of the access application; and through the analysis of historical data, screens out abnormal targets, reduces the amount of calculation, and improves the operation efficiency and network resource utilization.

Description

Technical Field

[0001] The invention belongs to the field of virtual machines, relates to network resource allocation technology, and specifically is a virtual machine network resource allocation system. Background Art

[0002] The demand for network resources by virtual machines changes dynamically, but the current resource allocation mechanism will lead to unreasonable network resources for virtual machines, resulting in excess network resources for some virtual machines.

[0003] For example, Chinese patent CN103001953B discloses a method and device for allocating network resources for a virtual machine, the method comprising: a first physical host receives a virtual machine resource request message sent by a user, the virtual machine resource request message is used to request allocation of network resources for at least one virtual machine; if the virtual machine resource request message requests allocation of network resources for one virtual machine, the first physical host allocates private VLAN resources to the user in the private VLAN resource pool of the first physical host, the private VLAN resources are used to isolate virtual machines in a layer 2 network; if the virtual machine resource request message requests allocation of network resources for at least two virtual machines, the first physical host allocates community VLAN resources to the user in the community VLAN resource pool of the first physical host, the community VLAN resources are used to connect virtual machines in a layer 2 network.

[0004] For example, Chinese patent CN109614229B relates to a virtual network resource allocation system based on software definition, Chinese patent CN109412865B relates to a virtual network resource allocation method, and so on, all of which disclose a network resource allocation technology.

[0005] However, the above-mentioned prior arts fail to allocate network resources based on the characteristics of the target object itself. Summary of the invention

[0006] The purpose of the present invention is to provide a virtual machine network resource allocation system.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] A virtual machine network resource allocation system includes: a single acquisition unit: which is used to acquire the habit data of a target object, calculate the required computing power range of the target object according to the habit data, and upload the habit data and the required computing power range to a single analysis unit; a single analysis unit: which performs demand analysis on the habit data of the target object, acquires the priority of the target object, and transmits the priority to a processor; a resource time control unit: which monitors the real-time computing power of the virtual machine and communicates with the processor; an abnormality identification unit: which performs abnormality analysis on non-target objects according to the habit data of the target object, and divides the non-target objects into abnormal target objects and normal target objects; the processor allocates resources to normal target objects according to the priority, the required computing power range and the real-time computing power; and an additional notification unit: which generates an abnormality warning after receiving the abnormal target object.

[0009] Furthermore, the single item acquisition unit is also used to acquire historical data of the request object, and divide the request target into target objects and non-target objects according to the historical data, and the following algorithm is executed when dividing the objects:

[0010] S1: retrieve the historical data corresponding to the requested object from the repository;

[0011] S2: If there is corresponding historical data, it is the target object, otherwise proceed to S3;

[0012] S3: Preprocess the request object:

[0013] Get the IP address of the request object;

[0014] Perform association analysis on the request object according to the target analysis rule to obtain the association degree Gg between the request object and the historical data stored in the repository;

[0015] If the correlation Gg≤G1, the requested object is a non-target object and G1 is the preset value;

[0016] Otherwise, the request object is the target object.

[0017] Furthermore, the degree of association Gg is obtained as follows:

[0018] G001: Get the estimated transmission path of the requested object and extract the IP address of the transmission destination;

[0019] G002: retrieve the IP address corresponding to the IP address of the transmission destination from the storage library and mark it as a historical IP address;

[0020] G003: Obtain the network resource usage corresponding to each scheduling of the historical IP address in the past six months, calculate the average network resource usage for a single use, and mark it as the average usage value Jz;

[0021] G004: The single-item acquisition unit retrieves the estimated network resource occupancy value Yz of the request object;

[0022] G005: Calculate the difference value Cz, Cz = |Jz-Yz|;

[0023] G006: If Cz ≥ C, then the correlation Among them, 1<α<2;

[0024] Otherwise, the correlation Among them, Zz is the average value of the single-use network resource usage corresponding to all IP addresses in the storage library, 1<α<β<2.

[0025] Furthermore, the method for the single item acquisition unit to estimate the expected network resource occupancy value Yz is:

[0026] Get the data transmission amount corresponding to the request object and mark it as the pre-transmission amount Qy;

[0027] Retrieve the reference target from the repository: mark the historical target object corresponding to the transmission volume Cy that satisfies |Cy-Qy|≤Q1 as the reference target; Cy is the transmission volume corresponding to the target object stored in the repository, where Q1 is the preset value;

[0028] Retrieve the network resource occupancy value Qzi corresponding to each reference target i from the storage library, and calculate the reference network resource occupancy value Qcc, i = 1, 2, 3, ..., n;

[0029] Estimated network resource usage If there is no transmission amount Cy that satisfies |Cy-Qy|≤Q1, then Qcc=0;

[0030] Among them, Qcs is the average network resource usage corresponding to all target objects stored in the repository.

[0031] Furthermore, the habit data includes historical transmission efficiency and network resource occupancy value corresponding to each target object j; when the single item acquisition unit calculates the required computing power range of the target object according to the habit data, the following algorithm is executed:

[0032] According to the historical transmission efficiency, the transmission time Ty is estimated;

[0033] Obtain the network resource occupancy value Qzj corresponding to each target object j stored in the repository;

[0034] The required computing power range is:

[0035] to

[0036] Among them, Qzmax is the maximum network resource occupancy value corresponding to all target objects stored in the repository.

[0037] Further: The method for estimating the transmission time is:

[0038] Obtain the data transmission efficiency corresponding to each target object j stored in the repository, marked as Xj, j = 1, 2, 3, ..., m;

[0039] Get the estimated network resource occupancy value Yz corresponding to the request target;

[0040] Estimated transfer time

[0041] Furthermore, the habit data also includes the network resource occupation time corresponding to the target object and the number of transmission nodes corresponding to the target object. The process of the single item analysis unit obtaining the priority of the target object is:

[0042] Retrieving habit data of the historical target object identical to the target object from the repository as proposed reference habit data;

[0043] According to the proposed reference habit data, the maximum network resource occupation time Tmax in the corresponding historical target object is obtained, and the average network resource occupation time Tzz corresponding to all historical target objects according to the habit data is obtained;

[0044] Obtain the request time of the target object, and number each target object in sequence according to the request time in the numbering rule of 0, 1, 0, 1, 0, 1… as the request target timing weight, and mark the request target timing weight as Sx;

[0045] Retrieve the number D of transmission nodes corresponding to the historical target object identical to the target object from the repository;

[0046] The priorities of the target objects are:

[0047] in, Express Rounded to the nearest integer, 0.432, 0.352, and 0.216 are the preset weights.

[0048] Furthermore, the resource time control unit calculates the real-time remaining computing power of the virtual machine in real time based on the real-time computing power of the virtual machine obtained through monitoring, takes the maximum required computing power corresponding to the required computing power range of the target object, and the processor allocates network resources to the target object in sequence according to the priority.

[0049] Furthermore, the abnormality identification unit performs abnormality analysis based on the habit data of the non-target object combined with the target object in the following process:

[0050] Obtain the average value of network resource usage corresponding to each target object stored in the repository;

[0051] Obtain the estimated network resource occupancy value and estimated transmission time corresponding to the non-target object;

[0052] If the estimated network resource usage value - the average network resource usage value > F1, and the estimated transmission time > F2, the non-target object is an abnormal target object;

[0053] Otherwise, the non-target object is a normal target object.

[0054] Furthermore, the notification unit generates an abnormal warning after receiving the abnormal target object, and displays it through the display unit. The administrator deletes the abnormal target object or modifies the abnormal target object into a normal target object through the management unit.

[0055] Beneficial effects of the present invention:

[0056] The present invention obtains the habit data of the target object through a single acquisition unit, calculates the required computing power range of the target object according to the habit data, and obtains the priority of the target object by performing demand analysis on the habit data of the target object; the resource time control unit monitors the real-time computing power of the virtual machine and communicates with the processor; the abnormal identification unit performs abnormal analysis on the non-target object according to the habit data of the target object, and divides the non-target object into abnormal target objects and normal target objects; the processor allocates resources to the normal target object according to the priority, the required computing power range and the real-time computing power; allocates network resource priority to each target object to ensure the throughput of the target object and the service quality of the access application; and through the analysis of historical data, screens out abnormal targets, reduces the amount of calculation, and improves the operation efficiency and network resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0058] Figure 1 A system block diagram of a virtual machine network resource allocation system of the present invention;

[0059] Figure 2 This is a flowchart of virtual machine network resource allocation according to the present invention. DETAILED DESCRIPTION

[0060] like Figure 1-2As shown, a virtual machine network resource allocation system includes: a single acquisition unit: which is used to acquire the habit data of the target object, calculate the required computing power range of the target object according to the habit data, and upload the habit data and the required computing power range to the single analysis unit; the single analysis unit: which performs demand analysis on the habit data of the target object, acquires the priority of the target object, and transmits the priority to the processor; a resource time control unit: which monitors the real-time computing power of the virtual machine and is connected to the processor in communication;

[0061] As an embodiment provided by the present invention, preferably, the habit data includes historical transmission efficiency and network resource occupancy value corresponding to each target object j; when the single item acquisition unit calculates the required computing power range of the target object according to the habit data, the following algorithm is executed:

[0062] According to the historical transmission efficiency, the transmission time Ty is estimated;

[0063] Obtain the network resource occupancy value Qzj corresponding to each target object j stored in the repository;

[0064] The required computing power range is:

[0065] to

[0066] Among them, Qzmax is the maximum network resource occupancy value corresponding to all target objects stored in the repository.

[0067] Abnormal identification unit: It performs abnormal analysis on non-target objects based on the habit data of the target objects, and divides the non-target objects into abnormal target objects and normal target objects; the processor allocates resources to normal target objects based on priority, required computing power range and real-time computing power; additional notification unit: It generates an abnormal warning after receiving the abnormal target object.

[0068] As an embodiment provided by the present invention, preferably, the resource timing control unit calculates the real-time remaining computing power of the virtual machine in real time based on the real-time computing power of the virtual machine obtained by monitoring, takes the maximum required computing power corresponding to the required computing power range of the target object, and the processor allocates network resources to the target object in sequence according to the priority.

[0069] As an embodiment provided by the present invention, preferably, the process of the abnormality identification unit performing abnormality analysis based on the habit data of the non-target object combined with the target object is:

[0070] Obtain the average value of network resource usage corresponding to each target object stored in the repository;

[0071] Obtain the estimated network resource occupancy value and estimated transmission time corresponding to the non-target object;

[0072] If the estimated network resource usage value - the average network resource usage value > F1, and the estimated transmission time > F2, the non-target object is an abnormal target object;

[0073] Otherwise, the non-target object is a normal target object.

[0074] As an embodiment provided by the present invention, preferably, the notification unit generates an abnormal warning after receiving the abnormal target object, and displays it through the display unit, and the administrator deletes the abnormal target object or modifies the abnormal target object into a normal target object through the management unit.

[0075] As an embodiment provided by the present invention, preferably, the single item acquisition unit is also used to acquire historical data of the request object, and divide the request target into target objects and non-target objects according to the historical data, and the following algorithm is executed when dividing the objects:

[0076] S1: retrieve the historical data corresponding to the requested object from the repository;

[0077] S2: If there is corresponding historical data, it is the target object, otherwise proceed to S3;

[0078] S3: Preprocess the request object:

[0079] Get the IP address of the request object;

[0080] According to the target parsing rule, the request object is associated and parsed to obtain the degree of association Gg between the request object and the historical data stored in the repository; as an embodiment provided by the present invention, preferably, the degree of association Gg is obtained in the following manner:

[0081] G001: Get the estimated transmission path of the requested object and extract the IP address of the transmission destination;

[0082] G002: retrieve the IP address corresponding to the IP address of the transmission destination from the storage library and mark it as a historical IP address;

[0083] G003: Obtain the network resource usage corresponding to each scheduling of the historical IP address in the past six months, calculate the average network resource usage for a single use, and mark it as the average usage value Jz;

[0084] G004: The single-item acquisition unit retrieves the estimated network resource occupancy value Yz of the request object;

[0085] G005: Calculate the difference value Cz, Cz = |Jz-Yz|;

[0086] G006: If Cz ≥ C, then the correlation Among them, 1<α<2;

[0087] Otherwise, the correlation Where Zz is the average single-use network resource usage corresponding to all IP addresses in the storage repository, 1<α<β<2;

[0088] If the correlation Gg≤G1, the requested object is a non-target object and G1 is the preset value;

[0089] Otherwise, the request object is the target object.

[0090] As an embodiment provided by the present invention, preferably, the method for the single item acquisition unit to estimate the expected network resource occupancy value Yz is:

[0091] Get the data transmission amount corresponding to the request object and mark it as the pre-transmission amount Qy;

[0092] Retrieve the reference target from the repository: mark the historical target object corresponding to the transmission volume Cy that satisfies |Cy-Qy|≤Q1 as the reference target; Cy is the transmission volume corresponding to the target object stored in the repository, where Q1 is the preset value;

[0093] Retrieve the network resource occupancy value Qzi corresponding to each reference target i from the storage library, and calculate the reference network resource occupancy value Qcc, i = 1, 2, 3, ..., n;

[0094] Estimated network resource usage If there is no transmission amount Cy that satisfies |Cy-Qy|≤Q1, then Qcc=0;

[0095] Among them, Qcs is the average network resource usage corresponding to all target objects stored in the repository.

[0096] As an embodiment provided by the present invention, preferably, the method for estimating the transmission time is:

[0097] Obtain the data transmission efficiency corresponding to each target object j stored in the repository, marked as Xj, j = 1, 2, 3, ..., m;

[0098] Get the estimated network resource occupancy value Yz corresponding to the request target;

[0099] Estimated transfer time

[0100] As an embodiment provided by the present invention, preferably, the habit data also includes the network resource occupation time corresponding to the target object and the number of transmission nodes corresponding to the target object, and the process of the single item analysis unit obtaining the priority of the target object is:

[0101] Retrieving habit data of the historical target object identical to the target object from the repository as proposed reference habit data;

[0102] According to the proposed reference habit data, the maximum network resource occupation time Tmax in the corresponding historical target object is obtained, and the average network resource occupation time Tzz corresponding to all historical target objects according to the habit data is obtained;

[0103] Obtain the request time of the target object, and number each target object in sequence according to the request time in the numbering rule of 0, 1, 0, 1, 0, 1… as the request target timing weight, and mark the request target timing weight as Sx;

[0104] Retrieve the number D of transmission nodes corresponding to the historical target object identical to the target object from the repository;

[0105] The priorities of the target objects are:

[0106] in, Express Rounded to the nearest integer, 0.432, 0.352, and 0.216 are the preset weights.

[0107] A virtual machine network resource allocation system obtains habit data of a target object through a single acquisition unit, calculates a required computing power range of the target object according to the habit data, and obtains the priority of the target object by performing a demand analysis on the habit data of the target object; a resource time control unit monitors the real-time computing power of the virtual machine and communicates with a processor; an abnormality identification unit performs abnormality analysis on non-target objects according to the habit data of the target object, and divides the non-target objects into abnormal target objects and normal target objects; the processor allocates resources to normal target objects according to the priority, the required computing power range and the real-time computing power; allocates network resource priority to each target object to ensure the throughput of the target object and the service quality of the access application; and through the analysis of historical data, screens out abnormal targets, reduces the amount of calculation, and improves the operation efficiency and network resource utilization.

[0108] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the claims, they should all fall within the protection scope of the present invention.

Claims

1. A virtual machine network resource allocation system, characterized in that: include: Single acquisition unit: used to acquire habit data of target object, calculate required computing power range of target object according to habit data, and upload habit data and required computing power range to single analysis unit; Single analysis unit: performs demand analysis on habit data of target object, acquires priority of target object, and transmits priority to processor; Resource time control unit: monitors real-time computing power of virtual machine and communicates with processor; Abnormal identification unit: performs abnormal analysis on non-target object according to habit data of target object, and divides non-target object into abnormal target object and normal target object; processor allocates resources to normal target object according to priority, required computing power range and real-time computing power; Additional notification unit: generates abnormal warning after receiving abnormal target object; The single item acquisition unit is also used to acquire historical data of the request object, and divide the request target into target objects and non-target objects according to the historical data. The following algorithm is executed when dividing the objects: S1: retrieve the historical data corresponding to the request object from the repository; S2: if the corresponding historical data exists, it is the target object, otherwise proceed to S3; S3: preprocess the request object: obtain the IP address of the request object; perform association analysis on the request object according to the target analysis rule, and obtain the association degree Gg between the request object and the historical data stored in the repository; if the association degree Gg≤G1, the request object is a non-target object, and G1 is a preset value; otherwise, the request object is a target object; the association degree Gg is obtained as follows: G001: obtain the expected transmission path of the request object and extract the IP address of the transmission end point; G002: retrieve the IP address corresponding to the IP address of the transmission end point from the repository and mark it as the historical IP address; G003: obtain the network resource occupancy corresponding to each scheduling of the historical IP address in the past six months, calculate the average network resource occupancy used for a single time, and mark it as the occupancy average Jz; G004: the single item acquisition unit retrieves the expected network resource occupancy value Yz of the request object; G005: calculate the difference value Cz, ; G006: If Cz ≥ C, then the correlation , where 1< <2; otherwise, the correlation , where Zz is the average single-use network resource usage corresponding to all IP addresses in the storage library, 1< < <2; The method for estimating the estimated network resource occupancy value Yz by the single item acquisition unit is: obtain the data transmission volume corresponding to the request object and mark it as the estimated transmission volume Qy; retrieve the reference target from the repository: satisfy The historical target object corresponding to the transmission volume Cy is marked as the reference target; Cy is the transmission volume corresponding to the target object stored in the repository, where Q1 is the preset value; the network resource occupancy value Qzi corresponding to each reference target i is retrieved from the repository, and the reference network resource occupancy value Qcc is calculated. ; i=1, 2, 3, ..., n; estimated network resource usage value If there is no satisfying The transmission volume Cy, then Qcc=0; wherein Qcs is the average value of network resource occupancy corresponding to all target objects stored in the repository; the habit data includes historical transmission efficiency and the network resource occupancy value corresponding to each target object j; When the single item acquisition unit calculates the required computing power range of the target object according to the habit data, the following algorithm is executed: based on the historical transmission efficiency, the transmission time Ty is estimated; the network resource occupancy value Qzj corresponding to each target object j stored in the repository is obtained; the required computing power range is: to ; Where Qzmax is the maximum network resource occupancy value corresponding to all target objects stored in the repository; The habit data also includes the network resource occupation time corresponding to the target object and the number of transmission nodes corresponding to the target object. The process of the single item analysis unit obtaining the priority of the target object is as follows: retrieve the habit data of the historical target object identical to the target object from the repository as the proposed reference habit data; obtain the maximum network resource occupation time Tmax in the corresponding historical target object according to the proposed reference habit data, and obtain the average network resource occupation time Tzz corresponding to all historical target objects according to the habit data; obtain the request time of the target object, and number each target object in sequence according to the request time according to the numbering rule of 0, 1, 0, 1, 0, 1... as the request target timing weight, and mark the request target timing weight as Sx; retrieve the number of transmission nodes D corresponding to the historical target object identical to the target object from the repository; the priority of the target object is: ;in, Express Rounded to the nearest integer, 0.432, 0.352, and 0.216 are the preset weights.

2. A virtual machine network resource allocation system according to claim 1, characterized in that: The method to estimate the transfer time is: Get the data transmission efficiency corresponding to each target object j stored in the repository, marked as Xj, j=1, 2, 3, ..., m; get the estimated network resource occupancy value Yz corresponding to the request target; Estimated transfer time .

3. A virtual machine network resource allocation system according to claim 1, characterized in that: The resource time control unit calculates the real-time remaining computing power of the virtual machine in real time based on the real-time computing power of the virtual machine obtained through monitoring, takes the maximum required computing power corresponding to the required computing power range of the target object, and the processor allocates network resources to the target object in sequence according to the priority.

4. A virtual machine network resource allocation system according to claim 1, characterized in that: The process of the abnormal identification unit performing abnormal analysis based on the habit data of the non-target object combined with the target object is as follows: Get the average network resource occupancy value corresponding to each target object stored in the repository; get the estimated network resource occupancy value and estimated transmission time corresponding to the non-target object; if the estimated network resource occupancy value - the average network resource occupancy value > F1, and the estimated transmission time > F2, the non-target object is an abnormal target object; otherwise, the non-target object is a normal target object, where F1 and F2 are preset values.

5. A virtual machine network resource allocation system according to claim 3, characterized in that: The notification unit generates an abnormal warning after receiving the abnormal target object, and displays it through the display unit. The administrator deletes the abnormal target object or modifies the abnormal target object into a normal target object through the management unit.

Images