Linq 查询的性能问题 - 应该归咎于导航属性？

Question

我有一个 Linq 查询花费的时间比它的等效 SQL 长得多（Linq 大约 5-6 秒，SQL 大约 50 毫秒）。显然翻译出了点问题，我什至知道是什么，但我不知道如何解决。查询如下：

from data in DbContext.CoacheeData
group data by new { data.User.CompanyId, data.Date } into g
select new GroupedDayData
{
    Id = g.FirstOrDefault().User.CompanyId == null ? "none" : g.FirstOrDefault().User.CompanyId,
    Name = g.FirstOrDefault().User.Company.Name == null ? "none" : g.FirstOrDefault().User.Company.Name,
    Date = g.FirstOrDefault().Date,
    Steps = g.Sum(x => x.Steps),
    Distance = g.Sum(x => x.Distance),
    CaloriesBurned = g.Sum(x => x.CaloriesBurned),
    LightlyActiveMinutes = g.Sum(x => x.LightlyActiveMinutes),
    FairlyActiveMinutes = g.Sum(x => x.FairlyActiveMinutes),
    VeryActiveMinutes = g.Sum(x => x.VeryActiveMinutes),
};

我怀疑在我的 group by 中使用导航属性 User 会导致为 CoacheeData 中的每条记录执行子查询。通过手动加入用户和公司表，我可以将效率提高大约 1.5 秒，如下所示：

from data in DbContext.CoacheeData
join user in DbContext.Users on data.UserId equals user.Id
join company in DbContext.Companies on user.CompanyId equals company.Id
group data by new { user.CompanyId, company.Name, data.Date } into g
select new GroupedDayData
{
    Id = g.FirstOrDefault().User.CompanyId == null ? "none" : g.FirstOrDefault().User.CompanyId,
    Name = g.FirstOrDefault().User.Company.Name == null ? "none" : g.FirstOrDefault().User.Company.Name,
    Date = g.FirstOrDefault().Date,
    Steps = g.Sum(x => x.Steps),
    Distance = g.Sum(x => x.Distance),
    CaloriesBurned = g.Sum(x => x.CaloriesBurned),
    LightlyActiveMinutes = g.Sum(x => x.LightlyActiveMinutes),
    FairlyActiveMinutes = g.Sum(x => x.FairlyActiveMinutes),
    VeryActiveMinutes = g.Sum(x => x.VeryActiveMinutes),
};

但我的选择看起来仍然相同，也许那里的导航属性导致了类似的问题？不过这看起来很奇怪，因为我之前在选择 Linq 查询时使用了 navigatio 属性而没有问题。无论如何，我不确定我的怀疑是否正确，如果正确，为什么会发生以及如何解决。

Answer 1

我怀疑获取 CompanyId、CompanyName 和 Date 的表达式过于复杂，Linq To Sql 无法在适当的 Sql 中进行翻译.

1) 尝试删除选择 CompanyId、CompanyName 和 Date 的部分，看看它是否变得更快；

var result =
    from data in DbContext.CoacheeData
    join user in DbContext.Users on data.UserId equals user.Id
    join company in DbContext.Companies on user.CompanyId equals company.Id
    group data by new { CompanyId = user.CompanyId, CompanyName = company.Name, Date = data.Date } into g
    select new GroupedDayData
    {
        Steps = g.Sum(x => x.Steps),
        Distance = g.Sum(x => x.Distance),
        CaloriesBurned = g.Sum(x => x.CaloriesBurned),
        LightlyActiveMinutes = g.Sum(x => x.LightlyActiveMinutes),
        FairlyActiveMinutes = g.Sum(x => x.FairlyActiveMinutes),
        VeryActiveMinutes = g.Sum(x => x.VeryActiveMinutes),
    };

2) 如果有帮助，请将您的表达式替换为 g.Key.CompanyId、g.Key.CompanyName 和 g.Key.Date 之类的内容

var result =
    from data in DbContext.CoacheeData
    join user in DbContext.Users on data.UserId equals user.Id
    join company in DbContext.Companies on user.CompanyId equals company.Id
    group data by new { CompanyId = user.CompanyId, CompanyName = company.Name, Date = data.Date } into g
    select new GroupedDayData
    {
        Id = g.Key.CompanyId,
        Name = g.Key.CompanyName,
        Date = g.Key.Date,
        Steps = g.Sum(x => x.Steps),
        Distance = g.Sum(x => x.Distance),
        CaloriesBurned = g.Sum(x => x.CaloriesBurned),
        LightlyActiveMinutes = g.Sum(x => x.LightlyActiveMinutes),
        FairlyActiveMinutes = g.Sum(x => x.FairlyActiveMinutes),
        VeryActiveMinutes = g.Sum(x => x.VeryActiveMinutes),
    };

3) 如果这有帮助，请添加空检查，您应该会得到相同的结果，但速度更快；

var result =
    from data in DbContext.CoacheeData
    join user in DbContext.Users on data.UserId equals user.Id
    join company in DbContext.Companies on user.CompanyId equals company.Id
    group data by new { CompanyId = user.CompanyId, CompanyName = company.Name, Date = data.Date } into g
    select new GroupedDayData
    {
        Id = g.Key.CompanyId ?? "None",
        Name = g.Key.CompanyName ?? "None",
        Date = g.Key.Date,
        Steps = g.Sum(x => x.Steps),
        Distance = g.Sum(x => x.Distance),
        CaloriesBurned = g.Sum(x => x.CaloriesBurned),
        LightlyActiveMinutes = g.Sum(x => x.LightlyActiveMinutes),
        FairlyActiveMinutes = g.Sum(x => x.FairlyActiveMinutes),
        VeryActiveMinutes = g.Sum(x => x.VeryActiveMinutes),
    };