EF 剝坑

 

一、簡單 count 會生成沒必要要的嵌套

var xs = (from x in dbContext.db_API_Operationallog where x.id<1  select 1 ).Count();

結果：

SELECT 
[GroupBy1].[A1] AS [C1]
FROM ( SELECT 
    COUNT(1) AS [A1]
    FROM [dbo].[db_API_Operationallog] AS [Extent1]
    WHERE [Extent1].[id] < 1
)  AS [GroupBy1]

二、查詢某些列的時候會整個模型查詢

    context.db_API_Operationallog.FirstOrDefault(p => p.Id == postId).Hits;
    或者：
    context.db_API_Operationallog.Find(postId).Hits;

會把整個post表數據都查出而後再內存裏面找了hits這個列

優化：

    context.db_API_Operationallog.Where(p => p.Id == postId).Select(p => p.Hits).FirstOrDefault();

三、不要輕易的把數據所有加載到內存

有時候一個不當心在查詢的sql build 裏面加了個tolist(), toArray()等，這種真正的執行了查詢，這樣在開發環境每每表的數據比較少，程序運行比較快，可是一到線上環境數據量比較大的狀況下就會出現內存爆滿的問題，這個問題相對來講比較隱蔽，因此開發的時候必定要當心。

四、IQueryable, IEnumerable

IEnumerable 執行的where 先是走內存，在走內存查詢

    public IEnumerable<db_API_Operationallog> GetAllPost()
    {
            return context.Post;
    }

    int id = 2000;
    var log = GetAllPost().Where(s => s.id <id).ToList();

Sql Server Profiler 抓到的信息

 SELECT 
[Extent1].[id] AS [id], 
[Extent1].[uid] AS [uid], 
[Extent1].[types] AS [types], 
[Extent1].[events] AS [events], 
[Extent1].[more] AS [more], 
[Extent1].[money] AS [money], 
[Extent1].[lastmoney] AS [lastmoney], 
[Extent1].[nowmoney] AS [nowmoney], 
[Extent1].[bak] AS [bak], 
[Extent1].[times] AS [times]
FROM [dbo].[db_API_Operationallog] AS [Extent1]

把上面的 IEnumerable 換成 IQueryable

exec sp_executesql N'SELECT 
[Extent1].[id] AS [id], 
[Extent1].[uid] AS [uid], 
[Extent1].[types] AS [types], 
[Extent1].[events] AS [events], 
[Extent1].[more] AS [more], 
[Extent1].[money] AS [money], 
[Extent1].[lastmoney] AS [lastmoney], 
[Extent1].[nowmoney] AS [nowmoney], 
[Extent1].[bak] AS [bak], 
[Extent1].[times] AS [times]
FROM [dbo].[db_API_Operationallog] AS [Extent1]
WHERE [Extent1].[id] < @p__linq__0',N'@p__linq__0 int',@p__linq__0=2000

這個坑是比較一樣也是比較隱蔽的注意下

五、使用 NoTracking 下降 狀態開銷

  dbContext.db_API_Operationallog.Where(s => s.id < id).AsNoTracking().ToList();

總結

平時寫EF的時候必定要把腳步停下了 多使用 Sql Server Profiler 琢磨下本身的代碼