测试Linq GroupBy DateTime时间3种方式的性能表现

先说结论：

按日期分组时，优先考虑用字符串方式对日期分组

var result = query.GroupBy(
		x => x.Time.ToString("yyyy/MM/dd"),
		(key, models) => new Output
		{
		    Date = key,
		    Count = models.Count()
		});

---

1.Linq GroupBy 匿名类型：

代码:

按日期分组并且计数

var result = query.GroupBy(
		x => new { x.Time.Date, x.Time.Date.Hour },
		(key, models) => new Output
		{
			Date = key.Date.ToString("yyyy/MM/dd HH"),
			Count = models.Count()
		})
		.ToList();

结果：

耗时大约836ms

---

2.Linq GroupBy dynamic：

代码:

//根据参数生成不同分组方式
Func<DashboardModel, dynamic> keySelector = null;
string dateFormat = null;
if (input.GroupByType == "days")
{
    keySelector = x => new { x.Time.Date };
    dateFormat = "yyyy/MM/dd";
}
else
{
    keySelector = x => new { x.Time.Date, x.Time.Date.Hour };
    dateFormat = "yyyy/MM/dd HH";
}
// 按日期分组并且计数
var result = query.GroupBy(
		keySelector,
		(key, models) => new Output
		{
		    Date = key.Date.ToString(dateFormat),
		    Count = models.Count()
		})
		.ToList();

结果：

耗时大约6s，性能明显下降

---

3.Linq GroupBy string：

代码:

//根据参数生成不同分组方式
string dateFormat = null;
if (input.GroupByType == "days")
{
    dateFormat = "yyyy/MM/dd";
}
else
{
    dateFormat = "yyyy/MM/dd HH";
}
}
// 按日期分组并且计数
var result = query.GroupBy(
		x => x.Time.ToString(dateFormat),
		(key, models) => new Output
		{
		    Date = key,
		    Count = models.Count()
		})
		.ToList();

结果：

耗时大约794ms，性能明显提升

---