多核环境下内存数据库并发调度技术优化研究

Multicore Environment Memory Database Technology on the Concurrent Scheduling Optimization

摘要:

对多核环境下内存数据进行并发调度,可以减少计算机宕机次数和数据切换时时间,提高数据并发调度精度,增加数据操作平稳性;当前的内存数据并发调度方法是利用PrebuiltTrigger对内存数据进行并发调度,在调度过程中,没有设定具体的内存数据调度目标,导致内存数据库中的数据因此错乱无序,存在数据并发调度精度低的问题;为此,提出一种基于Linux的多核环境下内存数据并发调度优化方法;该方法首先采用IACT算法对影响调度进行的数据和内存数据库中相似或重复数据进行清洗,然后以清洗的数据为基础,利用启发式算法对其进行数据特征选取,依据多属性决策理论对内存数据并发调度的最优路径属性权重集合进行计算,以其结果为依据,计算调度最优路径的偏差值,最后利用最小偏差值,建立调度最优路径线性规划模型,对每条调度路径的综合决策属性值进行排序,由此得到调度的最优路径,完成对多核环境下内存数据的并发调度;实验结果证明,所提方法可以对多核环境下内存数据进行高效率地并发调度,提高了数据调度精度,增加了内存数据的可循环利用性,为低开销的内存数据调度提供了支撑。

Abstract:

To multicore chips to concurrent scheduling, memory data can reduce the number of computer downtime and data when switching time, improve the accuracy of data concurrent scheduling, increase the smoothness of data operation. The current memory data concurrent scheduling method is to use PrebuiltTrigger to concurrent scheduling of data memory, in the process of scheduling, memory data scheduling goal is set, lead to memory data in the database disorder disorderly, therefore, has the problem of low data concurrent scheduling precision. For this, put forward a multi-core environment based on Linux memory data concurrent scheduling optimization method. This method firstly adopts IACT algorithm affect scheduling and memory of data in the database of similar or repeated data cleaning, and then on the basis of data cleaning, using heuristic algorithm for feature selection, data on the basis of the theory of multiple attribute decision making memory data concurrent scheduling of the optimal path through the calculation of the attribute weights are set based on the results, calculation of dispatching optimal path deviation, the use of the minimum deviation, scheduling the optimal path to the linear programming model is established, and the path to each of the scheduling of comprehensive decision attribute value to sort, the resulting scheduling optimal path, complete the memory data concurrent scheduling in multi-core environment. The experimental results show that the proposed method can memory multicore environment data efficiently concurrent scheduling, improved the precision of data scheduling, increased the memory data of recycled, provides low overhead of memory data scheduling with support.