我有一个cron-job方法,它根据用户的特色故事构建用户的故事提要,跟随类别并跟踪用户。
最终Feed以正确的顺序添加到下面的数据库表中:
UserFeed表:
Uid StoryListIds
1 3,23,45,3,6,234,.......
2 3,23,45,6,87,44,.......
3 3,23,45,32,4,62 .......
方法如下,并包含注释 代码:
public void ConstructUserFeed()
{
try
{
//get all user ids
var userIds = userBL.GetIds();
//get all featured stories
var featStories = storyBL.GetFeaturedStories();
if (userIds != null && userIds.Count > 0)
{
foreach (var userId in userIds)
{
//integer List to store the stories ids in correct order
List<int> storyIdsLst = new List<int>();
//integer List for excluding duplicates
List<int> exceptStoryIds = new List<int>();
//user feed List to store the stories
var userFeed = new List<UserFeedDTO>();
//string to store the id list so we can add it later in db
string storyIds = "";
if (featStories != null && featStories.Count > 0)
{
foreach (var featStory in featStories)
{
//first add all the ids of featured stories except own user stories, ordered by date
if (featStory.authorId != userId)
{
storyIdsLst.Add(featStory.id);
exceptStoryIds.Add(featStory.id);
}
}
}
//get user's followed categories ids
var followedCategoryIds = userCategoriesBL.GetFollowedCategoryIds(userId);
if (followedCategoryIds != null && followedCategoryIds.Count > 0)
{
foreach (var categoryId in followedCategoryIds)
{
//get the user's 5 latest stories for every followed category
//except own stories and previous stories
var storiesByCateg = storyBL.GetByCategory(5, categoryId, userId, exceptStoryIds);
if (storiesByCateg != null && storiesByCateg.Count > 0)
{
foreach (var storyByCateg in storiesByCateg)
{
userFeed.Add(storyByCateg);
exceptStoryIds.Add(storyByCateg.id);
}
}
}
}
//get user's followed users ids
var followedUserIds = userFollowersBL.GetFollowedUserIds(userId);
if (followedUserIds != null && followedUserIds.Count > 0)
{
foreach (var followedId in followedUserIds)
{
//get the user's 5 latest stories for every followed user
//except own stories and previous stories
var storiesByFollowedUsers = storyBL.GetByFollowedUser(5, followedId, userId, exceptStoryIds);
if (storiesByFollowedUsers != null && storiesByFollowedUsers.Count > 0)
{
foreach (var storyByFollowedUsers in storiesByFollowedUsers)
{
userFeed.Add(storyByFollowedUsers);
}
}
}
}
// order the stories by date
userFeed = userFeed.OrderByDescending(story => story.dateAdded).ToList();
if (userFeed != null && userFeed.Count > 0)
{
foreach (var story in userFeed)
{
//add the story ids after the featured story ids
storyIdsLst.Add(story.id);
}
}
//comma separated list of story ids as string so we can store it in db
storyIds = string.Join(",", storyIdsLst.Select(n => n.ToString()).ToArray());
//create the UserFeed model
UserFeed userFeedModel = new UserFeed();
userFeedModel.userId = userId;
userFeedModel.storyListId = storyIds;
userFeedModel.lastUpdateTime = DateTime.Now;
userFeedBL.AddOrUpdateUserFeed(userFeedModel);
}
uof.Save();
}
}
catch (Exception ex)
{
Console.WriteLine("Error occurred in processing job. Error : {0}", ex.Message);
}
}
上述方法需要约35秒才能完成30个用户 问:如何改进代码和性能?
答案 0 :(得分:1)
很难准确地说出导致它执行得如此缓慢的原因,因为我建议使用SQL Server Profiler来分析您的问题。检查您的查询是否被正确询问,并且没有做任何不必要的事情。
之后,我会考虑提出更少的问题。由于您在循环中执行此操作,因此可以从更少,但更重的问题中受益。例如(假设userFollowersBL。*正在查询数据库):
class myClass
{
public function methodReturnsTrue()
{
return true;
}
}
我认为签名是这样的:
var followedCategoryIds = userCategoriesBL.GetFollowedCategoryIds(userId);
考虑将其更改为:
IEnumerable<int> GetFollowedCategoryIds(userId);
然后,在开始使用foreach之前,您将在内存中找到包含userid和每个followCategoryIds的字典。这样,您就可以在一个查询中发送IDictionary<int, IEnumerable<int>> GetFollowedCategoryIds(IEnumerable<int> userIds);
的所有结果。 可能可以提高性能,而不是30次。同样适用于userBL.GetIds()
。
现在,您已使用约58次减少了对您的数据库的查询次数。
userFollowersBL.GetFollowedUserIds(userId)
但老实说,这主要是猜测,可能会不时发生变化。根据我的经验,您往往会从一次请求更多数据中受益,而不是多次使用相同的数据。