使 Linq 查询并行运行

Question

我可能有 40 个 linq 查询，我想并行执行它们。 查询之间没有依赖性，因此并行运行每个查询应该没问题。我想知道如何使下面的代码“更快”/“更好”。

我的两个基本查询和联合：

public override void Filter()
{
    var general = (from z in ctx.Interactions
                   where z.ActivityDate <= EndDate && z.ActivityDate >= StartDate &&
                         z.Indepth == false && z.Type == InteractionTypes.General
                   select new { Entries = z.EntryCount }).Sum(x => x.Entries).GetValueOrDefault();

    var Indepth = (from z in ctx.Interactions
                   join program in ctx.Programs on z.Program.Id equals program.Id
                   where z.ActivityDate <= EndDate && z.ActivityDate >= StartDate && z.Indepth == true && program.Reportable
                   select z).Count();

    ...Remaining 38 queries....  

    AddQuarterlyInfo("# of General Inquiries", "1.1", general);                 
    AddQuarterlyInfo("# of In-Depth Counselling and Information Services Interviews", "1.2", Indepth);
    ...Union remaining 38 queries...
}

Answer 1

如果您使用实体框架 6，则可以通过执行以下操作来使用异步查询：

var blogs = (from b in db.Blogs 
             orderby b.Name 
             select b).ToListAsync(); 

var blogs2 = (from b in db.Blogs 
             orderby b.Name 
             select b).ToListAsync(); 

var result = await Task.WhenAll(new [] {blogs, blogs2}));

//this is reached when all queries are completed. 

Answer 2

这对我有用，尽管如果没有相同的数据源，我无法谈论性能。

    int general = 0;
    int indepth = 0;
    List<Task> queryTasks = new List<Task>
        {
            new Task(() =>
            {
                general = (from a in ctx.Interactions
                    select new {}).Count();
            }),
            new Task(() =>
            {
                indepth = (from a in ctx.Interactions
                    select new {}).Count();
            })
        };
    Parallel.ForEach(queryTasks, q => q.RunSynchronously());

    await Task.WhenAll(queryTasks);

    AddQuarterlyInfo("# of General Inquiries", "1.1", general);                 
    AddQuarterlyInfo("# of In-Depth Counselling and Information Services Interviews", "1.2", indepth);