Summary of the invention
The object of the invention is to be achieved through the following technical solutions.
According to the embodiment of the present invention, propose a kind of resource management system based on cloud computing system, described system comprises: CPU (central processing unit), multiple data acquiring unit, acquisition data pre-processing unit, data summarization TU Trunk Unit and resource adjustment unit; Wherein,
Described multiple data acquiring unit, arranges respectively and runs on each distributed computer in described resource management system, obtains for the instruction according to CPU (central processing unit) the data message running on each analog machine of distributed computer;
Described acquisition data pre-processing unit, for carrying out pre-service to the data message obtained, and is sent to data summarization TU Trunk Unit by pretreated data message;
Described data summarization TU Trunk Unit, gathers through pretreated data message for receiving, and the data message gathered is sent to resource adjustment unit;
Described resource adjustment unit, for receiving the data message that data summarization TU Trunk Unit transmits, the distribution of preset rules to system resource according to CPU (central processing unit) adjusts.
According to the embodiment of the present invention, described acquisition data pre-processing unit specifically comprises: the 3rd serial port circuit, data calamity are for circuit, data layout decision circuitry, standardized data pre-process circuit and nonstandardized technique data prediction circuit; Wherein,
The data that described 3rd serial port circuit is sent by first serial circuit for receiving data acquiring unit;
Described data calamity is used for carrying out the standby process of calamity to the data that the data acquiring unit received sends for circuit;
The data that described data layout decision circuitry is used for the data capture unit received sends carry out form judgement, and described form comprises standardized format and nonstandardized technique form;
Described standardized data pre-process circuit is used for carrying out pre-service to standardized format data, comprises and the multiple standards formatted data of acquisition is converted to unified formatted data;
Described nonstandardized technique data prediction circuit is used for carrying out pre-service to nonstandardized technique formatted data; It comprises the characteristic obtaining described nonstandardized technique form, and described characteristic is stored, and then memory address is sent to CPU (central processing unit);
Described 3rd serial port circuit is also for being sent to data summarization TU Trunk Unit by above-mentioned through pretreated data according to CPU (central processing unit) instruction.
According to the embodiment of the present invention, described resource adjustment unit comprises first resource regulon and Secondary resource regulon, described first resource regulon is used for regulating system operations amount, and described Secondary resource regulon is used for regulating system cloud gray model validity.
According to the embodiment of the present invention, described first resource regulon specifically comprises: system actual operation amount computing unit, system operations amount anticipation unit and resource adjustments performance element, wherein,
Described system actual operation amount computing unit is used for the combined data real time computation system actual operation amount sent according to data summarization TU Trunk Unit;
Described system operations amount anticipation unit is used for the anticipation performing future operations amount according to the system operations amount of reality;
Described resource adjustments performance element is used for the adjustment carrying out system resource according to system actual operation amount and operand anticipation result.
According to the embodiment of the present invention, described resource adjustments performance element comprises: operand Information Receiving Circuits, code reassignment decision-making circuit, the real-time distributor circuit of resource and new calculation resources executive circuit, wherein,
Described operand Information Receiving Circuits receives operand anticipation result from system actual operation amount computing unit receiving system actual operation amount information and from system operations amount anticipation unit, and sends to code reassignment decision-making circuit;
Described code reassignment decision-making circuit carries out resource re-allocation decision according to the system actual operation amount information received and anticipation result, and determining whether needs increase for cloud computing and reduce resource; And the decision information of Resource dynamic allocation is sent to the real-time distributor circuit of resource;
If it is the decision information increasing resource that the real-time distributor circuit of resource receives, then from simulation the Resources list, distributes a simulation resource to cloud computing, then the simulation resource information increased is sent to new calculation resources executive circuit; New calculation resources executive circuit by cloud computing deployment in the simulation resource newly increased;
If it is the decision information reducing resource that the real-time distributor circuit of resource receives, then notify that the real-time distributor circuit of resource reclaims the simulation resource of described cloud computing; The real-time distributor circuit of resource reclaims simulation resource, is about to corresponding simulation resource and joins in simulation the Resources list.
According to the embodiment of the present invention, described Secondary resource regulon comprises:
Information Receiving Circuits, for receiving the system data information that data summarization TU Trunk Unit sends;
First analog machine statistical circuit, for obtaining the analog machine number had in current cloud computing system according to system data information;
Second analog machine statistical circuit, for adding up the analog machine number in current cloud computing system on every platform distributed computer according to system data information, and arranges according to the analog machine number on described every platform distributed computer;
Threshold decision circuit, for judging the minimum threshold number n of required distributed computer according to the maximum analog machine capacity of platform distributed computer every in described cloud computing system and described analog machine number;
Target group determination circuit, for selecting n distributed computer that analog machine number is maximum as target group according to the rank results of the second analog machine statistical circuit;
Data-moving group determination circuit, the analog machine on the distributed computer that will remove outside described target group as data-moving group, and arranges according to the analog machine number of the distributed computer outside the described target group of removing;
Data-moving circuit, for selecting successively from less to more according to analog machine number on distributed computer in data-moving group, and the present day analog machine of selection is carried out corresponding with described target group, until in data-moving group all analog machines all corresponding with described target group after, realize analog machine data-moving successively according to the corresponding relation between described data-moving group and described target group.
Resource management system based on cloud computing system of the present invention, by arranging CPU (central processing unit), multiple data acquiring unit, obtaining data pre-processing unit, data summarization TU Trunk Unit and resource adjustment unit, described system achieves the accurate control of fine adjustment to the operand of whole cloud computing system and validity, enhance data calamity for level simultaneously, improve the operational efficiency of whole system.
Embodiment
Below with reference to accompanying drawings illustrative embodiments of the present disclosure is described in more detail.Although show illustrative embodiments of the present disclosure in accompanying drawing, however should be appreciated that can realize the disclosure in a variety of manners and not should limit by the embodiment of setting forth here.On the contrary, provide these embodiments to be in order to more thoroughly the disclosure can be understood, and complete for the scope of the present disclosure can be conveyed to those skilled in the art.
According to the embodiment of the present invention, a kind of resource management system based on cloud computing system is proposed, as shown in Figure 1, described system comprises: CPU (central processing unit), multiple data acquiring unit, acquisition data pre-processing unit, data summarization TU Trunk Unit and resource adjustment unit; Wherein,
Described multiple data acquiring unit, arranges respectively and runs on each distributed computer in described resource management system, obtains for the instruction according to CPU (central processing unit) the data message running on each analog machine of distributed computer;
Described acquisition data pre-processing unit, for carrying out pre-service to the data message obtained, and is sent to data summarization TU Trunk Unit by pretreated data message;
Described data summarization TU Trunk Unit, gathers through pretreated data message for receiving, and the data message gathered is sent to resource adjustment unit;
Described resource adjustment unit, for receiving the data message that data summarization TU Trunk Unit transmits, the distribution of preset rules to system resource according to CPU (central processing unit) adjusts.
According to the embodiment of the present invention, as shown in Figure 2, described data acquiring unit specifically comprises:
Data receiver circuit, obtains CPU (central processing unit) instruction for receiving, and is obtained the data message of the analog machine operated on distributed computer by analog machine steady arm, then the information of acquisition is sent to data redundancy deletes circuit;
Data redundancy deletes circuit, for receiving the instruction obtaining CPU (central processing unit), to the analog machine data message that obtained data receiver circuit transmits, usage data redundancy Pruning algorithm carries out redundancy and deletes process, and the information after redundancy being deleted is stored in local memory;
Local memory, operates on distributed computer, the analog machine data message after deleting for storage redundancy;
Message reference circuit, under the control of the central processing unit, reads analog machine data message from local memory, and by described information transmission in serial port circuit; And
First serial circuit, is responsible for the passage providing data communication, for safeguard and communications and data between data summarization TU Trunk Unit and CPU (central processing unit) transmits; Receive the analog machine data message of message reference circuit transmission, and these information are sent to data pre-processing unit.
According to the embodiment of the present invention, as shown in Figure 3, described acquisition data pre-processing unit specifically comprises: the 3rd serial port circuit, data calamity are for circuit, data layout decision circuitry, standardized data pre-process circuit and nonstandardized technique data prediction circuit; Wherein,
The data that described 3rd serial port circuit is sent by first serial circuit for receiving data acquiring unit;
Described data calamity is used for carrying out the standby process of calamity to the data that the data acquiring unit received sends for circuit;
The data that described data layout decision circuitry is used for the data capture unit received sends carry out form judgement, and described form comprises standardized format and nonstandardized technique form; Described standardized format comprises, and such as, can use the data of bivariate table structure representation, described nonstandardized technique form comprises, such as, and text, picture, audio frequency etc.;
Described standardized data pre-process circuit is used for carrying out pre-service to standardized format data, comprises and the multiple standards formatted data of acquisition is converted to unified formatted data;
Described nonstandardized technique data prediction circuit is used for carrying out pre-service to nonstandardized technique formatted data; It comprises the characteristic obtaining described nonstandardized technique form, and described characteristic is stored, and then memory address is sent to CPU (central processing unit);
Described 3rd serial port circuit is also for being sent to data summarization TU Trunk Unit by above-mentioned through pretreated data according to CPU (central processing unit) instruction.
According to the embodiment of the present invention, described data calamity is carried out calamity for circuit specifically comprise for process the data that the data acquiring unit received sends:
D1, described CPU (central processing unit) obtain system operations amount information, and tasks carrying progress and positional information, determine disaster recovery solution;
D2, described data calamity judge for circuit task input data and divide, and to determine that data are standby the need of calamity, and propose the standby application of calamity to CPU (central processing unit);
D3, returning according to the standby order of the calamity of CPU (central processing unit), calculate and utilize idle bandwidth, considers to compare resource consumption, calamity for factors such as times, and it is standby to carry out data calamity;
D4, return calamity for situation to CPU (central processing unit), more new data calamity is for information;
Wherein, system operations amount information spinner described in described D1 will comprise round-trip delay and packet loss, namely increase with round-trip delay increase and packet loss in present period network and weigh system operations amount situation, too much cause overall performance to decline to avoid taking Internet resources because calamity is standby; Described tasks carrying progress, position are executed the task according to current, the MAC Address of reception task, tasks carrying analog machine position, task executed time, Task Progress score information, with the maximum duration needed for the complete execution of anticipation task and calamity for memory location;
Wherein, standby the relating generally to of the data calamity described in described D3 utilizes idle bandwidth calamity standby, considers that resource consumption compares and compares factor with calamity for the time; According to the system operations amount information that D1 receives, analyze its previous system operand rule, consider analog machine position, resource behaviour in service, calculate network idle condition in following 1 minute further, and it is standby to carry out data calamity this period;
Wherein, calamity described in described D4 for situation refer to by the anticipation of system operations amount and disaster recovery solution success calamity standby after information: data calamity for circuit after the standby success of intermediate data calamity, CPU (central processing unit) is sent to the form of heartbeat message, described in notice CPU (central processing unit), data calamity is for success and memory location, for the invalid rear use of current data; CPU (central processing unit) receives the calamity of data calamity for circuit for after information, upgrades data positional information.
Wherein, in described D3, calamity compares for the time and refers to that data calamity preparedness need be shorter than task execution time for the time, namely CPU (central processing unit) receives data calamity and carrys out control data calamity for the return message of circuit and ensure that data calamity is for the time for speed and computing node monitor task implementation progress two aspect, if calamity exceedes task completion time for the time, then abandon calamity standby; The previous time that resource consumption compares mainly through the actual execution of CPU (central processing unit) logger task compares for resource consumption as calamity for the time as the calamity of task execution resource consumption and intermediate data, be greater than calamity for resource consumption as performed resource consumption, then abandon calamity standby.
According to the embodiment of the present invention, as shown in Figure 4, described data summarization TU Trunk Unit specifically comprises:
Program control I/O sequence circuit, for the MAC Address of the distributed computer that receiving sequence management circuit sends, and by described address stored in program control I/O sequence, then under packing processing circuit controls, the MAC Address in program control I/O sequence is transferred to packing treatment circuit;
Sequence management circuit, receives the MAC Address of the distributed computer that data acquiring unit sends from second serial circuit, and sends described MAC Address to program control I/O sequence circuit; When the MAC Address in program control I/O sequence is invalid, described circuit is responsible for deleting the invalid MAC Address in program control I/O sequence;
Packing treatment circuit, is responsible for obtaining MAC Address periodically from program control I/O sequence circuit, and processes described MAC Address, after being packaged as data transfer instruction, described instruction is sent to second serial circuit;
Second serial circuit, is transmitted to sequence management circuit for the MAC Address sent to data summarization TU Trunk Unit by data acquiring unit; Receive the data transfer instruction that the transmission of packing treatment circuit comes simultaneously, then according to the MAC Address in described data transfer instruction, described data transfer instruction is sent to the data acquiring unit of specifying;
Data summarization circuit, as the core of data summarization TU Trunk Unit, receives the data message of the analog machine that pretreated multiple data acquiring unit obtains, and after combining these information, gathering, is sent to resource adjustment unit.
According to the embodiment of the present invention, described data summarization TU Trunk Unit carrying out practically step comprises:
A1, first determine combined data cycle of data summarization TU Trunk Unit: set data acquisition span cycle length of each data acquiring unit and it sends the time span of data to data summarization TU Trunk Unit;
Distributed computer address is sent out MAC by first serial circuit and is delivered to data summarization TU Trunk Unit by A2, data acquiring unit, and data summarization TU Trunk Unit uses second serial circuit to add in program control I/O sequence circuit by described address by sequence management circuit;
The program control I/O sequence circuit of packing treatment circuit cycle access of A3, data summarization TU Trunk Unit, after obtaining distributed computer MAC Address wherein, described MAC Address is treated to a data transfer instruction, more described instruction is sent to data acquiring unit by second serial circuit; While packing treatment circuit obtains distributed computer MAC Address, data summarization circuit prepares the analog machine data message receiving data acquiring unit transmission;
After A4, data summarization TU Trunk Unit start to receive data, using the data of current reception as a data block stored in buffer memory, and give described data block one unique identifier, for showing that current what gather is the acquisition data of which data acquiring unit;
A5, because of arrange data transmission period span determine, when described data acquiring unit end data transmission, data block in its buffer memory is just directly sent to distributed memory by data summarization TU Trunk Unit, thus completes the data summarization of a data acquiring unit;
A6, packing treatment circuit start to access program control I/O sequence circuit, start the acquisition data gathering next data acquiring unit, namely return and perform steps A 2.
According to the embodiment of the present invention, described resource adjustment unit comprises first resource regulon and Secondary resource regulon, described first resource regulon is used for regulating system operations amount, and described Secondary resource regulon is used for regulating system cloud gray model validity; Wherein,
As shown in Figure 5, described first resource regulon specifically comprises: system actual operation amount computing unit, system operations amount anticipation unit and resource adjustments performance element, wherein,
Described system actual operation amount computing unit is used for the combined data real time computation system actual operation amount sent according to data summarization TU Trunk Unit;
Described system operations amount anticipation unit is used for the anticipation performing future operations amount according to the system operations amount of reality;
Described resource adjustments performance element is used for the adjustment carrying out system resource according to system actual operation amount and operand anticipation result.
According to the embodiment of the present invention, the anticipation that described system operations amount anticipation unit performs future operations amount according to the system operations amount of reality specifically comprises:
E1, at moment t, calling system actual operation amount computing unit calculates system actual operation amount O (t) of moment t;
E2, execution short period span anticipation: utilize O (t) value and past data to calculate α (t) and E (t);
E3, execution long period span anticipation: utilize O (t) and past data to calculate α
t(t) and E
t(t);
E4, comprehensive short period span anticipation and long period span anticipation: as t < T, export O (t), go to step E5; Otherwise, get E (t-1), E
t(t-T) the two maximal value or mean value are as the output of moment t;
E5, renewal past data, wait for moment t+1, go to step E1.
Described step e 2 specifically comprises:
Under E21, initial situation, namely during t=0, get E (t)=O (t), proceed to step e 3; Otherwise, proceed to step e 22;
Square difference of E22, last short period span anticipation is f [α (t)]=[E (t-1)-O (t)]
2formula E (t)=α (t) E (t-1)+[1-α (t)] O (t) is substituted into, launch, utilize the past data preserved, structural differences polynomial f [α (t)]={ [E (t-3)-O (t-2)] * α (t)
2+ [O (t-2)-O (t-1)] * α (t)+[O (t-1)-O (t)] }
2;
E23, to difference polynomial f [α (t)] differentiate, obtain differentiable functions f ' [α (t)], solve its all solution on [-1,1] interval; If difference polynomial expression does not exist or f ' [α (t)] is upper without separating in [-1,1], getting α (t)=-1, proceeding to step e 25; Otherwise because f [α (t)] number of times is no more than four times, therefore f ' [α (the t)] skill on [-1,1] is no more than 3, and disaggregation is designated as S, proceeds to step e 24;
E24, ask difference polynomial f [α (t)] to S ∪ the functional value of all elements in-1,1}, compare their size, choosing the argument value corresponding to minimum in them is α (t); If there is multiple minimum functional value, get that minimum independent variable as α (t);
E25, calculating E (t)=α (t) E (t-1)+[1-α (t)] O (t).
Described step e 3 specifically comprises:
Under E31, initial situation, namely during t < T, ET (t)=O (t), proceeds to step e 4; Otherwise, proceed to step C2;
Square difference of E32, last long period span anticipation is f
t[α
t(t)]=[E
t(t-T)-O (t)]
2, by the cycle popularizing form E of formula
t(t)=α
t(t) E
t(t-T)+[1-α
t(t)] O (t) substitution, launches, utilizes the past data preserved, structural differences polynomial expression
f
T[α
T(t)]={[E
T(t-3T)-O(t-2T)]*α
T(t)2+[O(t-2T)-O(t-T)]*α
T(t)+[O(t-T)-O(t)]}
2;
E33, to described difference polynomial f
t[α
t(t)] differentiate, obtain differentiable functions f
t' [α
t(t)], solve its all solution on [-1,1] interval; If difference polynomial expression does not exist or f
t' [α
t(t)] upper without separating in [-1,1], get α
tt ()=-1, proceeds to step e 35; Otherwise, due to f
t[α
t(t)] number of times is no more than four times, therefore f
t' [α
t(t)] skill on [-1,1] is no more than 3, and disaggregation is designated as S
t, proceed to step e 34;
E34, ask difference polynomial f
t[α
t(t)] to S
t{ functional value of all elements in-1,1}, compare their size, choose the argument value corresponding to minimum in them is α to ∪
t(t); If there is multiple minimum functional value, get that minimum independent variable as α
t(t);
E35, calculating E
t(t)=α
t(t) E
t(t-T)+[1-α
t(t)] O (t).
According to the embodiment of the present invention, described resource adjustments performance element comprises: operand Information Receiving Circuits, code reassignment decision-making circuit, the real-time distributor circuit of resource and new calculation resources executive circuit, wherein,
Described operand Information Receiving Circuits receives operand anticipation result from system actual operation amount computing unit receiving system actual operation amount information and from system operations amount anticipation unit, and sends to code reassignment decision-making circuit;
Described code reassignment decision-making circuit carries out resource re-allocation decision according to the system actual operation amount information received and anticipation result, and determining whether needs increase for cloud computing and reduce resource; And the decision information of Resource dynamic allocation is sent to the real-time distributor circuit of resource;
If it is the decision information increasing resource that the real-time distributor circuit of resource receives, then from simulation the Resources list, distributes a simulation resource to cloud computing, then the simulation resource information increased is sent to new calculation resources executive circuit; New calculation resources executive circuit by cloud computing deployment in the simulation resource newly increased;
If it is the decision information reducing resource that the real-time distributor circuit of resource receives, then notify that the real-time distributor circuit of resource reclaims the simulation resource of described cloud computing; The real-time distributor circuit of resource reclaims simulation resource, is about to corresponding simulation resource and joins in simulation the Resources list.
According to the embodiment of the present invention, as shown in Figure 6, described Secondary resource regulon comprises:
Information Receiving Circuits, for receiving the system data information that data summarization TU Trunk Unit sends;
First analog machine statistical circuit, for obtaining the analog machine number had in current cloud computing system according to system data information;
Second analog machine statistical circuit, for adding up the analog machine number in current cloud computing system on every platform distributed computer according to system data information, and arranges according to the analog machine number on described every platform distributed computer;
Threshold decision circuit, for judging the minimum threshold number n of required distributed computer according to the maximum analog machine capacity of platform distributed computer every in described cloud computing system and described analog machine number, described n is positive integer;
Target group determination circuit, for selecting n distributed computer that analog machine number is maximum as target group according to the rank results of the second analog machine statistical circuit;
Data-moving group determination circuit, the analog machine on the distributed computer that will remove outside described target group as data-moving group, and arranges according to the analog machine number of the distributed computer outside the described target group of removing;
Data-moving circuit, for selecting successively from less to more according to analog machine number on distributed computer in data-moving group, and the present day analog machine of selection is carried out corresponding with described target group, until in data-moving group all analog machines all corresponding with described target group after, realize analog machine data-moving successively according to the corresponding relation between described data-moving group and described target group.
According to the embodiment of the present invention, described data-moving circuit specifically comprises:
Data-moving cost counting circuit, corresponding for the present day analog machine distributed computer maximum with analog machine number in target group that will select, and calculate corresponding total data and move cost, and using described data-moving cost as minimum cost;
Limit circuit, for other distributed computer in target group described in limit, calculate corresponding running total according to moving cost, and current data is moved total cost and described minimum cost compares, if current data moves total cost be more than or equal to described minimum cost, then repeat limit circuit, otherwise current data is moved total cost as minimum cost, present day analog machine is corresponding with current distributed computer.
According to the embodiment of the present invention, the total cost of described data-moving passes through following formulae discovery,
Cost(V
i,P
h)=mig×D(V
i,P
h)+∑[W(V
i,V
j)×D′(P
h,V
j)]
Wherein, Cost (V
i, P
h) be present day analog machine V
idata-moving is to current distributed computer P
htotal data move cost, mig is constant, D (V
i, P
h) be present day analog machine V
ito current distributed computer P
htopological step number, W (V
i, V
j) be present day analog machine V
iwith other analog machine V in cloud computing system
jbetween communication flows, D ' (P
h, V
j) be present day analog machine V
idata-moving is to distributed computer P
hafter, distributed computer P
hother analog machine V in cloud computing system
jdistance.
The above; be only the present invention's preferably embodiment, but protection scope of the present invention is not limited thereto, is anyly familiar with those skilled in the art in the technical scope that the present invention discloses; the change that can expect easily or replacement, all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should described be as the criterion with the protection domain of claim.