Linq查询的性能问题-归咎于导航属性？

Question

我有一个Linq查询，它花的时间远远超过它的SQL等效查询（Linq大约5-6秒，SQL 50毫秒）。 显然翻译中出现了问题，我什至知道一个问题，但是我不知道如何解决。 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),
};

我怀疑在我的组中使用导航属性User导致对CoacheeData中的每个记录执行子查询。 通过手动连接用户表和公司表，我可以将效率提高约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),
};

但是我的选择看起来仍然相同，也许那里的导航属性引起了类似的问题？ 但是，这似乎很奇怪，因为在没有问题的情况下，我在选择Linq查询时就使用了导航属性。 无论如何，我不确定我的猜想是否正确，如果怀疑是正确的，为什么会发生以及如何解决。

Answer 1

我怀疑要获取CompanyId ， CompanyName和Date的表达式太复杂，并且Linq To Sql无法在正确的Sql中将其转换。

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

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，请使用类似g.Key.CompanyId ， g.Key.CompanyName和g.Key.Date表达式替换表达式

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）如果有帮助，请添加null检查，您将获得相同的结果，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),
    };