CN103559333A - Biological gene sequencing task model building method based on log - Google Patents

Abstract

The invention discloses a biological gene sequencing task model building method based on a log. The method includes that workday periodicity and holiday periodicity of tasks in submit time in a biological gene sequencing log are analyzed and extracted; a task parallelism degree has the heavy tailed distribution characteristic in distribution, and the heavy tailed distribution characteristic exists between the task parallelism degree and the task operation time; queue use rate of a task queue has the exponential distribution, normal distribution, gamma distribution and binomial distribution characteristics, and a non-linear relationship exists between the queue use rate and queue day-task arriving number expectation. By means of the method, the workday periodicity and the holiday periodicity of the task in the submit time are simulated, then the task parallelism degree and the task operation time are generated; finally the relation between arriving number expectations generates a task queue number. The built task model can be used for well analyzing the advantages and the disadvantages of a biological gene sequencing technology, and resource use rate of a high-performance environment is optimized.

Description

The task model construction method of the biological gene order-checking based on daily record

Technical field

The present invention relates to high-performance computing sector, particularly a kind of task model construction method of the biological gene order-checking based on daily record.

Background technology

Since watson and crick have found that in nineteen fifty-three after the double-spiral structure of DNA, new chapter has just been opened in the development of life science.For the biological gene sequencing technologies of DNA, become the basis of whole life science development.Therefore on the other hand, biological gene sequencing technologies needs the calculating of magnanimity and storage resources to check order fast, if computational resource and storage resources are dispatched irrational words, can cause the utilization factor of resource low, and then postpone the speed of gene sequencing.And biological gene order-checking daily record is the use record of biological gene sequencing technologies under high-performance computing environment, by analyzing every attribute (time of arrival of task of task in biological gene order-checking daily record, the concurrency of task, the working time of task etc.), can grasp well the situation that computational resource is used in biological gene order-checking.And by these task characteristics, build the task model with identical characteristics, be conducive to propose a kind of colony dispatching strategy for these characteristics, and then optimize the resource utilization of cluster.Under high performance environments, utilize the task characteristic in task daily record to be divided into two kinds: plasticity task and rigidity task.Plasticity task refers to that the degree of parallelism of task and the working time of task are variable, and rigidity task refers to that the degree of parallelism of task and the working time of task are changeless.The present invention carries out model construction mainly for the task characteristic of rigidity task, the research situation that therefore task model of following article rigidity task direction builds.

The modeling of carrying out for rigidity task load in early days mainly contains following four features: the concurrency of task (the CPU quantity that task is used), the working time of task, user repeat the task quantity of submission and interval time of arrival of task.First utilization index distributes interval time of arrival of the task of simulating, and by log analysis being simulated to the concurrency of task, then produces Probability p by the concurrency of task, then utilizes this probability and high-order exponential distribution to simulate the working time of task.It is to be noted that this method carrying out interval when simulation task time, the working day that not consideration task arrives periodically and off-day periodicity.The present invention considers this two kinds of cyclophysises simultaneously.And in nearest research, researchist is by analyzing open question in above-mentioned research, characteristic etc. diurnal periodicity such as task interval time of arrival, then analyzed the periodicity on working day that task arrives, 48 time slots will be divided into for one day, each time slot (1800s), according to its average number of tasks arriving, obtains the weight that is proportional to number of tasks, then adopts gamma to distribute to periodically simulating the working day of task.Researchist finds that the concurrency of task and Runtime have proportional relation simultaneously, and the logarithm of these two task features is gamma and distributes, so researchist utilizes above-mentioned information first to simulate the concurrency of task, then simulates the working time of task by the concurrency of task.But not the working day at analysis task interval time of arrival periodically and festivals or holidays periodicity, but these two kinds are mixed and have analyzed characteristic diurnal periodicity.The present invention has considered this two specific character simultaneously.

Although existing research has been carried out good task model to rigidity task, build, still have problem to be solved, for example task working day cyclophysis and festivals or holidays cyclophysis model construction.Therefore, the present invention on the basis of existing research for working day at task interval time of arrival periodically and festivals or holidays periodicity, Runtime, the queue operating characteristic of the degree of parallelism of task and task has proposed a kind of task model construction method for biological gene order-checking based on daily record.

Summary of the invention

The shortcoming that the object of the invention is to overcome prior art is with not enough, task interval time of arrival for the biological gene order-checking daily record gathering from actual environment, the working time of task, the concurrency of task, and the queue characteristic of task builds a kind of task model with these task characteristics.

Object of the present invention is achieved through the following technical solutions:

A task model construction method for the biological gene order-checking of daily record, comprises the following steps:

(1) DCModel module construction, main utilization index distribution and gamma distribute the periodicity on working day of the task of simulating, and simulate periodicity festivals or holidays of task by exponential distribution, finally by two exponential distribution, produce the time interval T of task;

(2) PRModel module construction, first utilize the gamma distribution simulation degree of parallelism P that goes out on missions, then utilize the feature that has heavy-tailed distribution in biological gene order-checking daily record between tasks in parallel degree and Runtime, utilize this feature of gamma fitting of distribution, then simulate R working time of task;

(3) QModel module construction, first utilize a pseudo-random function generator to classify to all queues that will generate, be divided into four class LOW, MIDDLE, SUBHIGH, HIGH, adopts respectively exponential distribution for these four kinds of different classification, normal distribution, gamma distribution and binomial distribution generate the utilization rate U of each queue _i, then utilize the utilization rate U of queue _iwith the expectation of queue day task arrival number

between the nonlinear relationship that exists, generate

then by an exponential distribution, produce M _i, finally utilize U _i, M _iand a pseudo-random function generator produces queue number.

Preferably, the model construction of DCModel in step (1), first the start time S of judgement input is working day or festivals or holidays, if select periodicity module on working day working day, generates one and meets the time interval T that working day, periodic task arrived; Otherwise, generate one and meet periodic task interval T time of arrival festivals or holidays, finally use (S+T) as the submission time of this task, and revise start time S=S+T.

Preferably, the mathematic(al) representation that DCModel module realizes is as follows:

$\mathrm{WorkDC}\⇒\left\{\begin{array}{cc}{x}_i~E\left(\stackrel{\‾}{x_i}\right)& 1\≤i\≤9\\ x_i~\mathrm{\Γ}({\mathrm{\α}}_i,{\mathrm{\β}}_i)& 10\≤i\≤24\end{array}\right.$

$\mathrm{WeekDC}\⇒y_i~E\left(\stackrel{\‾}{y_i}\right),1\≤i\≤24$

$t_i~\left\{\begin{array}{cc}E(3600/x_i)& 1\≤i\≤24\\ E(3600/y_i)& 1\≤i\≤24\end{array}\right.$

Wherein, in formula, the meaning of parameters is: x _irepresent the task number that workaday time slot i arrives; y _irepresent the task number of the time slot i arrival of festivals or holidays;

represent the number of tasks object expectation value that working day, time slot i arrived;

represent the number of tasks object expectation value that festivals or holidays, time slot i arrived; E represents exponential distribution; Γ represents that gamma distributes; t _irepresent the time interval that i time slot task arrives; α _iand β _irepresent scale parameter and form parameter that gamma distributes; WorkDC represents working day periodically, and WeekDC represents that festivals or holidays periodically.

Preferably, the model construction of PRModel in step (2), the submission time (S+T) that reception is generated by step (1) is as triggering, by pseudo-random function generator, select the interval at tasks in parallel degree P place, then utilize corresponding gamma distribution function to produce the degree of parallelism of task; And then utilize pseudo-random function generator to select the interval at Runtime R place, then utilize corresponding gamma distribution function and tasks in parallel degree P to produce R working time of task.

Preferably, the mathematic(al) representation that PRModel module realizes is as follows:

$P~\left\{\begin{array}{cc}U(b_1,b_1)& P=b_1\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{11},{\mathrm{\&beta;}}_{11})& b_2\&le;P\&le;b_3\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{12},{\mathrm{\&beta;}}_{12})& b_3<P\&le;b_4\end{array}\right.$

$R(\&le;R_{\mathrm{low}\_\mathrm{th}})~\left\{\begin{array}{cc}\mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{21},{\mathrm{\&beta;}}_{21})& P=b_1\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{22},{\mathrm{\&beta;}}_{22})& b_2\&le;P\&le;b_3\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{23},{\mathrm{\&beta;}}_{23})& b_3<P\&le;b_4\end{array}\right.$

$R(>R_{\mathrm{mid}\_\mathrm{th}})~\left\{\begin{array}{cc}\mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{41},{\mathrm{\&beta;}}_{41})& P=b_1\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{42},{\mathrm{\&beta;}}_{42})& b_2\&le;P\&le;b_3\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{43},{\mathrm{\&beta;}}_{43})& b_3<P\&le;b_4\end{array}\right.$

Wherein, the meaning of parameters: b in above-mentioned four formula ₁, b ₂, b ₃, b ₄represent to divide the boundary value of tasks in parallel degree, relevant with the distribution of tasks in parallel degree in biological gene order-checking daily record; α ₁₁and β ₁₁, α ₁₂and β ₁₂represent the parameter value that tasks in parallel degree obedience gamma distributes; α ₂₁and β ₂₁, α ₂₂and β ₂₂, α ₂₃and β ₂₃, α ₃₁and β ₃₁, α ₃₂and β ₃₂, α ₃₃and β ₃₃, α ₄₁and β ₄₁, α ₄₂and β ₄₂, α ₄₃and β ₄₃represent that tasks in parallel degree is with the parameter value of the gamma distribution relation between Runtime; R _{low_th}and R _{mid_th}the threshold value that represents Runtime; U represents that consistance distributes.

Preferably, the model construction of QModel in step (3), (S+T) generating in receiving step (1) and (2), P, R is as triggering, according to the distribution character of queue in biological gene order-checking daily record, simulate and generate the queue number Q of this task, finally complete the structure of overall task model, the task of generation has the submission time of task, the degree of parallelism of task, the working time of task, the queue number of task.

Preferably, the mathematic(al) representation of QModel:

$U_i~\left\{\begin{array}{cc}E\left({\stackrel{\&OverBar;}{U}}_i\right)& i\&Element;\mathrm{LOW}\\ N({\mathrm{\&mu;}}_i,{\mathrm{\&sigma;}}_i)& i\&Element;\mathrm{MIDDLE}\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_i,{\mathrm{\&beta;}}_i)& i\&Element;\mathrm{SUBHIGH}\\ B(0.2,1.0)& i\&Element;\mathrm{HIGH}\end{array}\right.$

${\stackrel{\&OverBar;}{M}}_i=\left\{\begin{array}{cc}{l}_i{U_i}^3+l_2{U_i}^2+l_3{U}_i+l_4& i\&Element;\mathrm{LOW}\\ m_1{U_i}^3+m_2{U_i}^2+m_3{U}_i+m_4& i\&Element;\mathrm{MIDDLE}\\ s_1{U_i}^3+s_2{U_i}^2+s_3{U}_i+s_4& i\&Element;\mathrm{SUBHIGH}\end{array}\right.$

${\stackrel{\&OverBar;}{M}}_i=\left\{\begin{array}{cc}{C}_{0.2}& U_i=0.2,i\&Element;\mathrm{HIGH}\\ C_{1.0}& U_i=1.0,i\&Element;\mathrm{HIGH}\end{array}\right.$

$M_i~E\left({\stackrel{\&OverBar;}{M}}_i\right),i\&Element;\mathrm{LOW}\&cup;\mathrm{MIDDLE}\&cup;\mathrm{SUBHIGH}\&cup;\mathrm{HIGH}$

Wherein, the meaning of parameters: LOW in above-mentioned four formula, MIDDLE, SUBHIGH, HIGH is the interval of dividing according to the size of the expectation value of the day task arrival number of each queue in biological gene order-checking daily record; N represents normal distribution, and B is binomial distribution (being different from the binomial distribution of theory of probability the inside), mainly for generating the queue utilization rate in HIGH interval; U _ithe queue utilization rate that represents queue i, queue utilization rate represents that queue has the number of days of task arrival and the ratio of total number of days;

represent that the day task of queue i arrives the expectation value of number; M _ithe day task that represents queue i reaches number; C _0.2and C _1.0for the day task of two different queue utilization rates in biological gene, to arrive the expectation value (because these two numerical value are larger, therefore carry out independent simulation) of number;

μ _i, σ _i, α _i, β _ican draw concrete numerical value by the biological gene daily record of checking order.）

The present invention has following advantage and effect with respect to prior art:

The working day that 1.DCModel has simulated task in biological gene order-checking daily record periodically and festivals or holidays periodically, and the task model in the past date periodicity of the task of consideration just does not distinguish working day and festivals or holidays.

2.PRModel checks order biological gene to connect the working time of task and the degree of parallelism of task in daily record, first produce the degree of parallelism of task, then utilize tasks in parallel degree with the relation of Runtime, utilize the working time of gamma distribution generation task.

3.QModel first simulate task in biological gene order-checking daily record and, then utilize this two queue number that simulated behavior is gone out on missions, and in existing research, do not consider the simulation of the queue number of task.

4. different from existing research, the present invention be directed to the model construction that biological gene order-checking daily record is carried out, therefore can be better for the analysis of biological gene sequencing technologies.

Accompanying drawing explanation

Fig. 1 is for the present invention is based on the frame diagram of the task model of biological gene order-checking daily record structure.

Embodiment

Below in conjunction with embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are not limited to this.

Embodiment

As shown in Figure 1, the method for the task model building based on biological gene order-checking daily record, comprises the following steps:

(1) structure of DCModel.In biological gene order-checking daily record interval time of arrival of task exist working day cyclophysis and festivals or holidays cyclophysis, so DCModel need can produce simultaneously have working day cyclophysis and festivals or holidays cyclophysis task daily record.First 24 time slots will be divided into for one day, corresponding one hour of each time slot, characteristic diurnal periodicity of biological gene order-checking daily record shows: the number that each time slot task arrives between time slot 1 and time slot 9 presents the process then raising that reduces gradually, and at time slot 10, to the number that between time slot 24, each time slot task arrives, presents the process that raises and then reduce gradually; At time slot 1, to the interior task of time slot 9, arrive number simultaneously and have exponential distribution relation with time slot, and to time slot 24, there is gamma distribution relation in time slot 10.

$\mathrm{WorkDC}\&DoubleRightArrow;\left\{\begin{array}{cc}{x}_i~E\left(\stackrel{\&OverBar;}{x_i}\right)& 1\&le;i\&le;9\\ x_i~\mathrm{\&Gamma;}({\mathrm{\&alpha;}}_i,{\mathrm{\&beta;}}_i)& 10\&le;i\&le;24\end{array}\right.$

$\mathrm{WeekDC}\&DoubleRightArrow;y_i~E\left(\stackrel{\&OverBar;}{y_i}\right),1\&le;i\&le;24$

$t_i~\left\{\begin{array}{cc}E(3600/x_i)& 1\&le;i\&le;24\\ E(3600/y_i)& 1\&le;i\&le;24\end{array}\right.$

For in biological gene order-checking daily record diurnal periodicity characteristic, the specific implementation of DCModel of the present invention is as above-mentioned three formula.The construction step of DCModel is as follows:

1) first by the biological gene daily record of checking order, can calculate in each time slot on working day

α _iand β _i, then calculate x _i;

2) can from checking order daily record, biological gene calculate equally each time slot festivals or holidays

then calculate y _i;

3) from existing research, easily know the time interval obeys index distribution that in cluster, task arrives, the time span of a time slot is 3600s simultaneously, therefore can utilize above-mentioned last formula, x _iand y _iproduce the time interval of working day and task arrival festivals or holidays, thereby complete the model construction of DCModel.

(2) structure of PRModel.Between the working time of task in biological gene order-checking daily record and the degree of parallelism of task, there is heavy-tailed distribution relation.But the data generated error of heavy-tailed distribution is larger, therefore first the present invention is divided into different tasks in parallel degree intervals by the degree of parallelism of task according to different sizes, then utilizes gamma to distribute the degree of parallelism of interior task between matching division back zone with the relation between Runtime.The mathematical relation expression formula that following four formula are PRModel.

$P~\left\{\begin{array}{cc}U(b_1,b_1)& P=b_1\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{11},{\mathrm{\&beta;}}_{11})& b_2\&le;P\&le;b_3\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{12},{\mathrm{\&beta;}}_{12})& b_3<P\&le;b_4\end{array}\right.$

$R(\&le;R_{\mathrm{low}\_\mathrm{th}})~\left\{\begin{array}{cc}\mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{21},{\mathrm{\&beta;}}_{21})& P=b_1\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{22},{\mathrm{\&beta;}}_{22})& b_2\&le;P\&le;b_3\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{23},{\mathrm{\&beta;}}_{23})& b_3<P\&le;b_4\end{array}\right.$

$R(>R_{\mathrm{mid}\_\mathrm{th}})~\left\{\begin{array}{cc}\mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{41},{\mathrm{\&beta;}}_{41})& P=b_1\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{42},{\mathrm{\&beta;}}_{42})& b_2\&le;P\&le;b_3\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{43},{\mathrm{\&beta;}}_{43})& b_3<P\&le;b_4\end{array}\right.$

The construction step of PRModel is as follows:

1) first in above-mentioned first formula, utilize biological gene order-checking daily record to calculate the shared probable value of task in different degree of parallelisms interval, utilize pseudo-random function generator to produce a probability, utilize this probable value to select corresponding degree of parallelism interval, then utilize corresponding distribution function to produce the degree of parallelism value P of task;

2) then utilize biological gene order-checking daily record, calculate the probable value of pdf (P, R), wherein pdf represents the probable value between the interval of different degree of parallelism P and the interval of different task R working time.Wherein P is by b ₁, b ₂, b ₃, b ₄be divided into three different degree of parallelisms interval, and R is by R _{low_th}, R _{mid_th}be divided into three intervals.Simultaneously, by step 1) drawn the interval at tasks in parallel degree place, and utilize biological gene order-checking daily record can draw the probable value between should three Runtime Zone Rs in tasks in parallel degree P interval, then utilize a pseudo-random function generator can select the interval of corresponding Runtime R.

3) finally utilize step 1) in the interval of P and the interval of Runtime R of the tasks in parallel degree that draws, and above-mentioned last three formula select suitable gamma distribution function to produce R working time of task.

(3) structure of QModel.In biological gene order-checking daily record, use many queues Task Scheduling Model, the utilization rate of each queue is not identical, and have certain regularity, so the present invention has built a kind of QModel model for biological gene order-checking journal queue by this regularity of research and analysis.Below four relationships that formula is QModel.

$U_i~\left\{\begin{array}{cc}E\left({\stackrel{\&OverBar;}{U}}_i\right)& i\&Element;\mathrm{LOW}\\ N({\mathrm{\&mu;}}_i,{\mathrm{\&sigma;}}_i)& i\&Element;\mathrm{MIDDLE}\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_i,{\mathrm{\&beta;}}_i)& i\&Element;\mathrm{SUBHIGH}\\ B(0.2,1.0)& i\&Element;\mathrm{HIGH}\end{array}\right.$

${\stackrel{\&OverBar;}{M}}_i=\left\{\begin{array}{cc}{l}_i{U_i}^3+l_2{U_i}^2+l_3{U}_i+l_4& i\&Element;\mathrm{LOW}\\ m_1{U_i}^3+m_2{U_i}^2+m_3{U}_i+m_4& i\&Element;\mathrm{MIDDLE}\\ s_1{U_i}^3+s_2{U_i}^2+s_3{U}_i+s_4& i\&Element;\mathrm{SUBHIGH}\end{array}\right.$

${\stackrel{\&OverBar;}{M}}_i=\left\{\begin{array}{cc}{C}_{0.2}& U_i=0.2,i\&Element;\mathrm{HIGH}\\ C_{1.0}& U_i=1.0,i\&Element;\mathrm{HIGH}\end{array}\right.$

$M_i~E\left({\stackrel{\&OverBar;}{M}}_i\right),i\&Element;\mathrm{LOW}\&cup;\mathrm{MIDDLE}\&cup;\mathrm{SUBHIGH}\&cup;\mathrm{HIGH}$

The specific implementation step of QModel is as follows:

1) first from biological gene order-checking daily record, calculate LOW, MIDDLE, SUBHIGH, tetra-shared ratios in interval of HIGH, then utilize the queue number that a pseudo-random function will generate to be distributed to randomly in these four intervals.Then calculate the U of each queue _iwith

finally calculate the M of each queue _i.

2) by above-mentioned first formula, calculate the interval at queue i place, and select suitable distribution function, utilize corresponding distribution function to produce the queue utilization rate U of queue i _i.Interval LOW in biological gene order-checking daily record, in MIDDLE and SUBHIGH, the U of queue _iwith queue

have nonlinear relationship, the present invention utilizes three three binomials to carry out U in these three intervals of matching _iwith

relation; And in interval HIGH, because queue is less, so queue utilization rate U _iwith

adopt one to one relation as shown in above-mentioned the 3rd formula.Then can from above-mentioned the 3rd and the 4th formula, select the U that queue i is corresponding _iwith relational expression obtain queue repeating step 2), until obtain the U of all queues _iwith

3) in biological gene order-checking daily record, the M of queue _idistribution be to obey expectation value to be

exponential distribution.Therefore can pass through above-mentioned the 4th formula, utilize queue i's simultaneously

obtain the M of queue _i.

4) finally utilize the U of each queue _iand M _igenerate queue number: first utilize a pseudorandom number generator to generate the numerical value between 0 and 1, by all U _ian interim list is put in the queue that is greater than this numerical value; Then by all queues in this interim list according to they M _isize criteriaization in interval [0,1], according to M _isize corresponding interval with ratio in interval [0,1]; Finally, recycle another pseudorandom number generator and produce the numerical value between 0 and 1, the queue number of the interval correspondence at numerical value place is the queue number of final generation.

Last comprehensive step (1) (2) (3) generates and has the job invocation time, the working time of task, the task of the degree of parallelism of task and the queue number of task.Repeat these three steps and can generate a plurality of task records, thereby be formed for the Performance Evaluation of biological gene sequencing technologies.

Above-described embodiment is preferably embodiment of the present invention; but embodiments of the present invention are not restricted to the described embodiments; other any do not deviate from change, the modification done under Spirit Essence of the present invention and principle, substitutes, combination, simplify; all should be equivalent substitute mode, within being included in protection scope of the present invention.

Claims (7)

1. a task model construction method for the order-checking of the biological gene based on daily record, is characterized in that, comprises the following steps:

(1) DCModel module construction, main utilization index distribution and gamma distribute the periodicity on working day of the task of simulating, and simulate periodicity festivals or holidays of task by exponential distribution, finally by two exponential distribution, produce the time interval T of task;

(2) PRModel module construction, first utilize the gamma distribution simulation degree of parallelism P that goes out on missions, then utilize the feature that has heavy-tailed distribution in biological gene order-checking daily record between tasks in parallel degree and Runtime, utilize this feature of gamma fitting of distribution, then simulate R working time of task;

(3) QModel module construction, first utilize a pseudo-random function generator to classify to all queues that will generate, be divided into four class LOW, MIDDLE, SUBHIGH, HIGH, adopts respectively exponential distribution for these four kinds of different classification, normal distribution, gamma distribution and binomial distribution generate the utilization rate U of each queue _i, then utilize the utilization rate U of queue _iwith the expectation of queue day task arrival number

between the nonlinear relationship that exists, generate

then by an exponential distribution, produce M _i, finally utilize U _i, M _iand a pseudo-random function generator produces queue number.

2. the task model construction method that the biological gene based on daily record according to claim 1 checks order, it is characterized in that, the model construction of DCModel in step (1), first the start time S of judgement input is working day or festivals or holidays, if select periodicity module on working day working day, generate one and meet the time interval T that working day, periodic task arrived; Otherwise, generate one and meet periodic task interval T time of arrival festivals or holidays, finally use (S+T) as the submission time of this task, and revise start time S=S+T.

3. the task model construction method of the biological gene order-checking based on daily record according to claim 2, is characterized in that, the mathematic(al) representation that DCModel module realizes is as follows:

$\mathrm{WorkDC}\&DoubleRightArrow;\left\{\begin{array}{cc}{x}_i~E\left(\stackrel{\&OverBar;}{x_i}\right)& 1\&le;i\&le;9\\ x_i~\mathrm{\&Gamma;}({\mathrm{\&alpha;}}_i,{\mathrm{\&beta;}}_i)& 10\&le;i\&le;24\end{array}\right.$

$\mathrm{WeekDC}\&DoubleRightArrow;y_i~E\left(\stackrel{\&OverBar;}{y_i}\right),1\&le;i\&le;24$

$t_i~\left\{\begin{array}{cc}E(3600/x_i)& 1\&le;i\&le;24\\ E(3600/y_i)& 1\&le;i\&le;24\end{array}\right.$

Wherein, in formula, the meaning of parameters is: x _irepresent the task number that workaday time slot i arrives; y _irepresent the task number of the time slot i arrival of festivals or holidays;

represent the number of tasks object expectation value that working day, time slot i arrived;

represent the number of tasks object expectation value that festivals or holidays, time slot i arrived; E represents exponential distribution; Γ represents that gamma distributes; t _irepresent the time interval that i time slot task arrives; α _iand β _irepresent scale parameter and form parameter that gamma distributes; WorkDC represents working day periodically, and WeekDC represents that festivals or holidays periodically.

4. the task model construction method that the biological gene based on daily record according to claim 1 checks order, it is characterized in that, the model construction of PRModel in step (2), the submission time (S+T) that reception is generated by step (1) is as triggering, by pseudo-random function generator, select the interval at tasks in parallel degree P place, then utilize corresponding gamma distribution function to produce the degree of parallelism of task; And then utilize pseudo-random function generator to select the interval at Runtime R place, then utilize corresponding gamma distribution function and tasks in parallel degree P to produce R working time of task.

5. the task model construction method of the biological gene order-checking based on daily record according to claim 4, is characterized in that, the mathematic(al) representation that PRModel module realizes is as follows:

$P~\left\{\begin{array}{cc}U(b_1,b_1)& P=b_1\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{11},{\mathrm{\&beta;}}_{11})& b_2\&le;P\&le;b_3\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{12},{\mathrm{\&beta;}}_{12})& b_3<P\&le;b_4\end{array}\right.$

$R(\&le;R_{\mathrm{low}\_\mathrm{th}})~\left\{\begin{array}{cc}\mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{21},{\mathrm{\&beta;}}_{21})& P=b_1\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{22},{\mathrm{\&beta;}}_{22})& b_2\&le;P\&le;b_3\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{23},{\mathrm{\&beta;}}_{23})& b_3<P\&le;b_4\end{array}\right.$

$R(>R_{\mathrm{mid}\_\mathrm{th}})~\left\{\begin{array}{cc}\mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{41},{\mathrm{\&beta;}}_{41})& P=b_1\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{42},{\mathrm{\&beta;}}_{42})& b_2\&le;P\&le;b_3\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_{43},{\mathrm{\&beta;}}_{43})& b_3<P\&le;b_4\end{array}\right.$

Wherein, the meaning of parameters: b in above-mentioned four formula ₁, b ₂, b ₃, b ₄represent to divide the boundary value of tasks in parallel degree, relevant with the distribution of tasks in parallel degree in biological gene order-checking daily record; α ₁₁and β ₁₁, α ₁₂and β ₁₂represent the parameter value that tasks in parallel degree obedience gamma distributes; α ₂₁and β ₂₁, α ₂₂and β ₂₂, α ₂₃and β ₂₃, α ₃₁and β ₃₁, α ₃₂and β ₃₂, α ₃₃and β ₃₃, α ₄₁and β ₄₁, α ₄₂and β ₄₂, α ₄₃and β ₄₃represent that tasks in parallel degree is with the parameter value of the gamma distribution relation between Runtime; R _{low_th}and R _{mid_th}the threshold value that represents Runtime; U represents that consistance distributes.

6. the task model construction method that the biological gene based on daily record according to claim 1 checks order, it is characterized in that, the model construction of QModel in step (3), (S+T) generating in receiving step (1) and (2), P, R is as triggering, according to the distribution character of queue in biological gene order-checking daily record, simulate and generate the queue number Q of this task, finally complete the structure of overall task model, generating of task has the submission time of task, the degree of parallelism of task, the working time of task, the queue number of task.

7. the task model construction method that the biological gene based on daily record according to claim 6 checks order, is characterized in that the mathematic(al) representation of QModel:

$U_i~\left\{\begin{array}{cc}E\left({\stackrel{\&OverBar;}{U}}_i\right)& i\&Element;\mathrm{LOW}\\ N({\mathrm{\&mu;}}_i,{\mathrm{\&sigma;}}_i)& i\&Element;\mathrm{MIDDLE}\\ \mathrm{\&Gamma;}({\mathrm{\&alpha;}}_i,{\mathrm{\&beta;}}_i)& i\&Element;\mathrm{SUBHIGH}\\ B(0.2,1.0)& i\&Element;\mathrm{HIGH}\end{array}\right.$

${\stackrel{\&OverBar;}{M}}_i=\left\{\begin{array}{cc}{l}_i{U_i}^3+l_2{U_i}^2+l_3{U}_i+l_4& i\&Element;\mathrm{LOW}\\ m_1{U_i}^3+m_2{U_i}^2+m_3{U}_i+m_4& i\&Element;\mathrm{MIDDLE}\\ s_1{U_i}^3+s_2{U_i}^2+s_3{U}_i+s_4& i\&Element;\mathrm{SUBHIGH}\end{array}\right.$

${\stackrel{\&OverBar;}{M}}_i=\left\{\begin{array}{cc}{C}_{0.2}& U_i=0.2,i\&Element;\mathrm{HIGH}\\ C_{1.0}& U_i=1.0,i\&Element;\mathrm{HIGH}\end{array}\right.$

$M_i~E\left({\stackrel{\&OverBar;}{M}}_i\right),i\&Element;\mathrm{LOW}\&cup;\mathrm{MIDDLE}\&cup;\mathrm{SUBHIGH}\&cup;\mathrm{HIGH}$

Wherein, the meaning of parameters: LOW in above-mentioned four formula, MIDDLE, SUBHIGH, HIGH is the interval of dividing according to the size of the expectation value of the day task arrival number of each queue in biological gene order-checking daily record; N represents normal distribution, and B is binomial distribution, mainly for generating the queue utilization rate in HIGH interval; U _ithe queue utilization rate that represents queue i, queue utilization rate represents that queue has the number of days of task arrival and the ratio of total number of days;

represent that the day task of queue i arrives the expectation value of number; M _ithe day task that represents queue i reaches number; C _0.2and C _1.0for the day task of two different queue utilization rates in biological gene, to arrive the expectation value of number;

μ _i, σ _i, α _i, β _ican draw concrete numerical value by the biological gene daily record of checking order.