Abstract:Aiming at the uneven distribution of resources caused by excessive discreteness of resources and large amount of data under the PaaS platform, an automatic allocation algorithm based on multiple regression is proposed to effectively solve the problem. Count all resource data samples, use the multiple regression algorithm to calculate the sample set with the smallest residual sum of squares in the observation sequence, and substitute it into the loss function to obtain the loss difference, obtain the minimum loss value, and find the optimal sample of the gradient value in the set according to the gradient descent law , using this sample as the allocation reference. Set up a set of resource data constraints for allocating the highest correlation with the reference sample value, and iteratively calculate until all the resources that meet the constraints are obtained. The qualified ones are allocated to one group, and the rest are allocated to another group. According to the time and bit distribution sequence of resources, discrete capture is continuously implemented to find residual resources, and then secondary allocation is implemented. Simulation experiments show that the PaaS platform resource throughput is increased after the proposed method is allocated, the consumption cost is reduced, the algorithm allocation time is low, and the overall practical ability is strong. |