实体框架6 .异步方法与同步方法的性能
本文关键字：同步方法 性能 异步方法 框架 实体 | 更新日期: 2023-09-27 18:16:15
我已经修改了一个Github项目来测试一些更多的实体框架相关的查询方式。
链接模型:
[Table("Player")]
public partial class Player
{
public int Id { get; set; }
[Required]
[StringLength(200)]
public string FirstName { get; set; }
[Required]
[StringLength(200)]
public string LastName { get; set; }
public DateTime DateOfBirth { get; set; }
public int TeamId { get; set; }
public virtual Team Team { get; set; }
}
[Table("Team")]
public partial class Team
{
public Team()
{
Players = new HashSet<Player>();
}
public int Id { get; set; }
[Required]
[StringLength(200)]
public string Name { get; set; }
public DateTime FoundingDate { get; set; }
public int SportId { get; set; }
public virtual ICollection<Player> Players { get; set; }
public virtual Sport Sport { get; set; }
}
[Table("Sport")]
public partial class Sport
{
public Sport()
{
Teams = new HashSet<Team>();
}
public int Id { get; set; }
[StringLength(100)]
public string Name { get; set; }
public virtual ICollection<Team> Teams { get; set; }
}
我要测试的是这些东西:
EF"同步"——FirstOrDefault()和ToList ()
context.Players.FirstOrDefault(x => x.Id == id);
context.Players.AsNoTracking().Where(x => x.TeamId == teamId).ToList();
context.Teams.AsNoTracking().Include(x => x.Players).Where(x => x.SportId == sportId).ToList();
EF "Async" - FirstOrDefaultAsync() and ToListAsync()
await context.Players.FirstOrDefaultAsync(x => x.Id == id);
await context.Players.AsNoTracking().Where(x => x.TeamId == teamId).ToListAsync();
await context.Teams.AsNoTracking().Include(x => x.Players).Where(x => x.SportId == sportId).ToListAsync();
EF "Sync" with Select into DTO对象
context.Players.Select(p => new PlayerDTO()
{
Id = p.Id,
FirstName = p.FirstName,
DateOfBirth = p.DateOfBirth,
LastName = p.LastName,
TeamId = p.TeamId
}).FirstOrDefault(x => x.Id == id);
context.Players.Where(x => x.TeamId == teamId).Select(p => new PlayerDTO()
{
Id = p.Id,
FirstName = p.FirstName,
DateOfBirth = p.DateOfBirth,
LastName = p.LastName,
TeamId = p.TeamId
}).ToList();
context.Teams.Where(x => x.SportId == sportId).Select(t => new TeamDTO()
{
Id = t.Id,
FoundingDate = t.FoundingDate,
Name = t.Name,
SportId = t.SportId,
Players = t.Players.Select(p => new PlayerDTO()
{
Id = p.Id,
FirstName = p.FirstName,
DateOfBirth = p.DateOfBirth,
LastName = p.LastName,
TeamId = p.TeamId
}).ToList()
}).ToList();
EF "Async" with Select into DTO对象
await context.Players.Select(p => new PlayerDTO()
{
Id = p.Id,
FirstName = p.FirstName,
DateOfBirth = p.DateOfBirth,
LastName = p.LastName,
TeamId = p.TeamId
}).FirstOrDefaultAsync(x => x.Id == id);
await context.Players.Where(x => x.TeamId == teamId).Select(p => new PlayerDTO()
{
Id = p.Id,
FirstName = p.FirstName,
DateOfBirth = p.DateOfBirth,
LastName = p.LastName,
TeamId = p.TeamId
}).ToListAsync();
await context.Teams.Where(x => x.SportId == sportId).Select(t => new TeamDTO()
{
Id = t.Id,
FoundingDate = t.FoundingDate,
Name = t.Name,
SportId = t.SportId,
Players = t.Players.Select(p => new PlayerDTO()
{
Id = p.Id,
FirstName = p.FirstName,
DateOfBirth = p.DateOfBirth,
LastName = p.LastName,
TeamId = p.TeamId
}).ToList()
}).ToListAsync();
我认为我的结果很麻烦。主要是因为微软正在大力推广EF的异步部分。
我的结果:(以毫秒为单位)
项目数量:8，球队数量:30，球员数量:100
EntityFrameworkAsyncDTO Results
Run # Player by ID Players per Team Teams per Sport
0 1,46 3,47 35,88
1 1,04 3 33
2 1,02 3,3 33,75
3 1,03 3 31,75
4 1,1 3,27 31,38
EntityFrameworkAsync Results
Run # Player by ID Players per Team Teams per Sport
0 1,17 3,53 57
1 1,01 3 48,62
2 0,99 3,03 47,88
3 1,02 3,07 51
4 1 3,03 48,88
EntityFrameworkDTO Results
Run # Player by ID Players per Team Teams per Sport
0 0,02 3 13,25
1 0,01 2,8 13,12
2 0,15 2,97 13,25
3 0,02 2,9 13,25
4 0,05 2,8 13,12
EntityFramework Results
Run # Player by ID Players per Team Teams per Sport
0 0,01 2,27 28,38
1 0,01 2,4 28,25
2 0 2,13 28,5
3 0,01 2,17 27,5
4 0,01 2,13 29
ADONET Results
Run # Player by ID Players per Team Teams per Sport
0 0 2,03 11,75
1 0 2 12,62
2 0 2 11,38
3 0 2 12,38
4 0 2 11,25
/EDIT我已经添加了ADO。
我的问题是:我做错了我的查询，因为加载时间是如此不同?
在我运行的真实世界系统中，我已经将所有异步DB调用更改为同步调用，并且我可以在我的统计数据中看到系统看到DB调用的速度增加了50%。这是一个100-120并发用户通过SignalR连接的系统，DB非常重
我认为这篇文章解释得很详细。
Async并不快，而且在普通的"同步"代码之上增加了很多开销。但是它允许在你浪费时间等待(例如网络连接的高延迟等)的情况下更好地使用资源
我同意其他贡献者的意见，如果您按照以下方式运行异步测试:
var tskPlayers = context.Players.FirstOrDefaultAsync(x => x.Id == id);
var tskPlayersNt = context.Players.AsNoTracking().Where(x => x.TeamId == teamId).ToListAsync();
var tskTeamsNt = context.Teams.AsNoTracking().Include(x => x.Players).Where(x => x.SportId == sportId).ToListAsync();
Task.WhenAll(tskPlayers, tskPlayersNt, tskTeamsNt);
异步方法意味着更多的IL代码，从而解释了一次性调用的差异。
如果你在你的方法中进行其他异步调用，你可以获得异步方法的好处。
另一篇有趣的文章:https://www.thereformedprogrammer.net/analysis-of-entity-framework-6-async-performance/