19、并行关系数据仓库设计与太阳物理大数据分析

并行关系数据仓库设计与太阳物理大数据分析

1 并行关系数据仓库查询分配算法

在并行关系数据仓库中，查询分配是一个关键环节。以下是查询分配算法的详细步骤：

Algorithm 1. Query Allocation( M node, Qj Query)
1: Let ListFrag the list of valid fragments for Qj.
2: Let NumberFrag the number of fragments in ListFrag;
3: Let NumberValideNode the number of valid nodes for the fragments in ListFrag;
4: Let ListSubQuery the sub-query list : /**each valid fragment will give rise to a
sub-query**/
5: Estimate SizeQ the number of Inputs/Output (IO) needed to execute Qj;
6: Calculate MPS be mean data processing of Qj ;
LB =
1
NumberV alideNode
j=1
1
jδ
× SizeQ
(3)
7: Sort ListFrag according to their size in descending order;
8: for i = 1 to NumberFrag do
9:
Get the valid nodes for the ith fragment in ListFrag a