很久以前写了一篇文章 .NET中使用Redis 介绍了如何安装Redis服务端,以及如何在.NET中调用Redis读取数据。本文简单介绍如何设计NoSQL数据库,以及如何使用Redis来存储对象。
和传统的关系型数据库不同,NoSQL大部分都是以键值对存储在内存中的,我们不能直接把RDBMS里面的一些做法直接移植到NoSQL中来,一个最主要的原因是,在NoSQL中缺少RDBMS中的一些诸如join ,union以及一些在关系型数据库中效率很高的执行语句,这些在NoSQL不能很好的支持,或者说效率低。
下文首先通过例子介绍在SQLServer中设计一个DB系统以及与NoSQL环境中设计一个DB的区别,最后演示如何在Redis中对数据进行读写操作。
假设我们要设计一个简单的博客系统,用户可以注册一个博客(Blog),然后可以在上面写文章(Post),文章可以分类(Category)以及添加标签(Tag),用户可以对文章进行评论(Comment)。
在该系统中,我们需要实现,如下基本功能:
如果在SQLServer中,相信很简单就可以设计出这样一个DB了。
database in SQLServer" src="/Upload/Images/2014112610/77330D5DA339BA22.png" alt="Blog database in SQLServer" width="654" height="406" border="0" />
在NoSQL环境中,我们不能直接将上面的结构搬进来,所以需要根据需求重新设计我们的模型。
在NoSQL环境下,所有的数据其实都是以key和value的形式存储在内存中的,value通常是序列化为字符串保存的。我们使用redis客户端的时候,可以直接将对象存储,这些客户端在内部实现上帮助我们进行了序列化。所以第一步就是需要定义实体模型:
首先来看User实体:
class="code">public class User { public User() { this.BlogIds = new List<long>(); } public long Id { get; set; } public string Name { get; set; } public List<long> BlogIds { get; set; } }
User实体中,包含了用户的Id,Name以及博客的Id。
然后Blog实体:
public class Blog { public Blog() { this.Tags = new List<string>(); this.BlogPostIds = new List<long>(); } public long Id { get; set; } public long UserId { get; set; } public string UserName { get; set; } public List<string> Tags { get; set; } public List<long> BlogPostIds { get; set; } }
包含了标签Tag,以及文章Id列表。
文章BolgPost实体:
public class BlogPost { public BlogPost() { this.Categories = new List<string>(); this.Tags = new List<string>(); this.Comments = new List<BlogPostComment>(); } public long Id { get; set; } public long BlogId { get; set; } public string Title { get; set; } public string Content { get; set; } public List<string> Categories { get; set; } public List<string> Tags { get; set; } public List<BlogPostComment> Comments { get; set; } }
包含了一篇文章的基本信息,如文章分类,文章标签,文章的评论。
最后看评论BlogPostComment实体:
public class BlogPostComment { public string Content { get; set; } public DateTime CreatedDate { get; set; } }
实体定义好了之后,我们就可以开始具体实现了。为了演示,这里通过单元测试的方式实现具体功能:
首先要把Redis的服务端启动起来,然后在工程中新建一个Redis客户端,之后的所有操作都通过这个客户端进行。
[TestFixture, Explicit, Category("Integration")] public class BlogPostExample { readonly RedisClient redis = new RedisClient("localhost"); [SetUp] public void OnBeforeEachTest() { redis.FlushAll(); InsertTestData(); } }
在单元测试的SetUp中,我们插入一些模拟数据,插入数据的方法为InsetTestData方法:
public void InsertTestData() { var redisUsers = redis.As<User>(); var redisBlogs = redis.As<Blog>(); var redisBlogPosts = redis.As<BlogPost>(); var yangUser = new User { Id = redisUsers.GetNextSequence(), Name = "Eric Yang" }; var zhangUser = new User { Id = redisUsers.GetNextSequence(), Name = "Fish Zhang" }; var yangBlog = new Blog { Id = redisBlogs.GetNextSequence(), UserId = yangUser.Id, UserName = yangUser.Name, Tags = new List<string> { "Architecture", ".NET", "Databases" }, }; var zhangBlog = new Blog { Id = redisBlogs.GetNextSequence(), UserId = zhangUser.Id, UserName = zhangUser.Name, Tags = new List<string> { "Architecture", ".NET", "Databases" }, }; var blogPosts = new List<BlogPost> { new BlogPost { Id = redisBlogPosts.GetNextSequence(), BlogId = yangBlog.Id, Title = "Memcache", Categories = new List<string> { "NoSQL", "DocumentDB" }, Tags = new List<string> {"Memcache", "NoSQL", "JSON", ".NET"} , Comments = new List<BlogPostComment> { new BlogPostComment { Content = "First Comment!", CreatedDate = DateTime.UtcNow,}, new BlogPostComment { Content = "Second Comment!", CreatedDate = DateTime.UtcNow,}, } }, new BlogPost { Id = redisBlogPosts.GetNextSequence(), BlogId = zhangBlog.Id, Title = "Redis", Categories = new List<string> { "NoSQL", "Cache" }, Tags = new List<string> {"Redis", "NoSQL", "Scalability", "Performance"}, Comments = new List<BlogPostComment> { new BlogPostComment { Content = "First Comment!", CreatedDate = DateTime.UtcNow,} } }, new BlogPost { Id = redisBlogPosts.GetNextSequence(), BlogId = yangBlog.Id, Title = "Cassandra", Categories = new List<string> { "NoSQL", "Cluster" }, Tags = new List<string> {"Cassandra", "NoSQL", "Scalability", "Hashing"}, Comments = new List<BlogPostComment> { new BlogPostComment { Content = "First Comment!", CreatedDate = DateTime.UtcNow,} } }, new BlogPost { Id = redisBlogPosts.GetNextSequence(), BlogId = zhangBlog.Id, Title = "Couch Db", Categories = new List<string> { "NoSQL", "DocumentDB" }, Tags = new List<string> {"CouchDb", "NoSQL", "JSON"}, Comments = new List<BlogPostComment> { new BlogPostComment {Content = "First Comment!", CreatedDate = DateTime.UtcNow,} } }, }; yangUser.BlogIds.Add(yangBlog.Id); yangBlog.BlogPostIds.AddRange(blogPosts.Where(x => x.BlogId == yangBlog.Id).Map(x => x.Id)); zhangUser.BlogIds.Add(zhangBlog.Id); zhangBlog.BlogPostIds.AddRange(blogPosts.Where(x => x.BlogId == zhangBlog.Id).Map(x => x.Id)); redisUsers.Store(yangUser); redisUsers.Store(zhangUser); redisBlogs.StoreAll(new[] { yangBlog, zhangBlog }); redisBlogPosts.StoreAll(blogPosts); }
在方法中,首先在Redis中创建了三个强类型的IRedisTypedClient类型的对象redisUsers,redisBlogs,redisBlogPosts来保存用户信息,博客信息,和文字信息。
var yangUser = new User { Id = redisUsers.GetNextSequence(), Name = "Eric Yang" };
在新建用户的时候,因为Id是自增字段,所以直接调用redisUsers这个client的GetNextSequence()方法就可以获得一个自增的Id。
创建完用户之后,接着创建博客信息:
var yangBlog = new Blog { Id = redisBlogs.GetNextSequence(), UserId = yangUser.Id, UserName = yangUser.Name, Tags = new List<string> { "Architecture", ".NET", "Databases" }, };
该博客有几个标签。
在接着创建该博客上发表的若干篇文章:
var blogPosts = new List<BlogPost> { new BlogPost { Id = redisBlogPosts.GetNextSequence(), BlogId = yangBlog.Id, Title = "Memcache", Categories = new List<string> { "NoSQL", "DocumentDB" }, Tags = new List<string> {"Memcache", "NoSQL", "JSON", ".NET"} , Comments = new List<BlogPostComment> { new BlogPostComment { Content = "First Comment!", CreatedDate = DateTime.UtcNow,}, new BlogPostComment { Content = "Second Comment!", CreatedDate = DateTime.UtcNow,}, } } }
每一篇文章都有分类和标签,以及评论。
然后需要给user的BlogsIds和blog的BlogPostIds赋值
yangUser.BlogIds.Add(yangBlog.Id);
yangBlog.BlogPostIds.AddRange(blogPosts.Where(x => x.BlogId == yangBlog.Id).Map(x => x.Id));
最后需要把这些信息保存到redis中。
//保存用户信息 redisUsers.Store(yangUser); redisUsers.Store(zhangUser); //保存博客信息 redisBlogs.StoreAll(new[] { yangBlog, zhangBlog }); //保存所有的文章信息 redisBlogPosts.StoreAll(blogPosts);
现在,利用Redis Desktop Manager,可以查看Reidis中存储的数据:
数据准备好了之后,可以实现前面列出的一系列方法了:
该方法在GetAllBlogs中,实现如下:
[Test] public void Show_a_list_of_blogs() { var redisBlogs = redis.As<Blog>(); var blogs = redisBlogs.GetAll(); blogs.PrintDump(); }
只需要调用GetAll方法即可获取内存中的所有指定类型的对象。
输出结果为:
[
{
Id: 1,
UserId: 1,
UserName: Eric Yang,
Tags:
[
Architecture,
.NET,
Databases
],
BlogPostIds:
[
1,
3
]
},
{
Id: 2,
UserId: 2,
UserName: Fish Zhang,
Tags:
[
Architecture,
.NET,
Databases
],
BlogPostIds:
[
2,
4
]
}
]
实现如下:
[Test] public void Show_a_list_of_recent_posts_and_comments() { //Get strongly-typed clients var redisBlogPosts = redis.As<BlogPost>(); var redisComments = redis.As<BlogPostComment>(); { //To keep this example let's pretend this is a new list of blog posts var newIncomingBlogPosts = redisBlogPosts.GetAll(); //Let's get back an IList<BlogPost> wrapper around a Redis server-side List. var recentPosts = redisBlogPosts.Lists["urn:BlogPost:RecentPosts"]; var recentComments = redisComments.Lists["urn:BlogPostComment:RecentComments"]; foreach (var newBlogPost in newIncomingBlogPosts) { //Prepend the new blog posts to the start of the 'RecentPosts' list recentPosts.Prepend(newBlogPost); //Prepend all the new blog post comments to the start of the 'RecentComments' list newBlogPost.Comments.ForEach(recentComments.Prepend); } //Make this a Rolling list by only keep the latest 3 posts and comments recentPosts.Trim(0, 2); recentComments.Trim(0, 2); //Print out the last 3 posts: recentPosts.GetAll().PrintDump(); recentComments.GetAll().PrintDump(); } }
方法中定义了两个key为urn:BlogPost:RecentPosts 和 urn:BlogPostComment:RecentComments的 List对象来保存最近发表的文章和评论:recentPosts.Prepend(newBlogPost)方法表示将新创建的文章插到recentPosts列表的最前面。
Trim方法表示仅保留n个在集合中。
显示博客的标签云方法如下:
[Test] public void Show_a_TagCloud() { //Get strongly-typed clients var redisBlogPosts = redis.As<BlogPost>(); var newIncomingBlogPosts = redisBlogPosts.GetAll(); foreach (var newBlogPost in newIncomingBlogPosts) { //For every tag in each new blog post, increment the number of times each Tag has occurred newBlogPost.Tags.ForEach(x => redis.IncrementItemInSortedSet("urn:TagCloud", x, 1)); } //Show top 5 most popular tags with their scores var tagCloud = redis.GetRangeWithScoresFromSortedSetDesc("urn:TagCloud", 0, 4); tagCloud.PrintDump(); }
显示标签云的实现,用到了redis中的SortedSet,IncrementItemInSortedSet表示如果有相同的话,值加一,GetRangeWithScoresFromSortedSetDesc方法,获取某一key的前5个对象。
显示所有的分类用到了Set对象。
[Test] public void Show_all_Categories() { var redisBlogPosts = redis.As<BlogPost>(); var blogPosts = redisBlogPosts.GetAll(); foreach (var blogPost in blogPosts) { blogPost.Categories.ForEach(x => redis.AddItemToSet("urn:Categories", x)); } var uniqueCategories = redis.GetAllItemsFromSet("urn:Categories"); uniqueCategories.PrintDump(); }
实现如下:
[Test] public void Show_post_and_all_comments() { //There is nothing special required here as since comments are Key Value Objects //they are stored and retrieved with the post var postId = 1; var redisBlogPosts = redis.As<BlogPost>(); var selectedBlogPost = redisBlogPosts.GetById(postId.ToString()); selectedBlogPost.PrintDump(); }
只需要把postId传进去就可以通过GetById的方法获取内存中的对象.
首先根据PostId获取BlogPost,然后在Comment属性中添加一个BlogPostComment对象,然后在保存改BlogPost.
[Test] public void Add_comment_to_existing_post() { var postId = 1; var redisBlogPosts = redis.As<BlogPost>(); var blogPost = redisBlogPosts.GetById(postId.ToString()); blogPost.Comments.Add( new BlogPostComment { Content = "Third Post!", CreatedDate = DateTime.UtcNow }); redisBlogPosts.Store(blogPost); var refreshBlogPost = redisBlogPosts.GetById(postId.ToString()); refreshBlogPost.PrintDump(); }
[Test] public void Show_all_Posts_for_the_DocumentDB_Category() { var redisBlogPosts = redis.As<BlogPost>(); var newIncomingBlogPosts = redisBlogPosts.GetAll(); foreach (var newBlogPost in newIncomingBlogPosts) { //For each post add it's Id into each of it's 'Cateogry > Posts' index newBlogPost.Categories.ForEach(x => redis.AddItemToSet("urn:Category:" + x, newBlogPost.Id.ToString())); } //Retrieve all the post ids for the category you want to view var documentDbPostIds = redis.GetAllItemsFromSet("urn:Category:DocumentDB"); //Make a batch call to retrieve all the posts containing the matching ids //(i.e. the DocumentDB Category posts) var documentDbPosts = redisBlogPosts.GetByIds(documentDbPostIds); documentDbPosts.PrintDump(); }
这里首先把所有的文章按照标签新建Set,把相同的分类的文章放到一个Set中,最后根据key即可查找到相应的集合。
本文利用一个简单的博客系统,简要介绍了如何利用Redis存储和获取复杂的数据。由于本文主要为了演示如何与Redis进行交互,所以实体设计的很简陋,没有按照DDD的思想进行设计,在某些设计方面没有遵循前文浅谈依赖注入中使用的原理和方法,后面会写文章对该系统进行重构以使之更加完善。
希望本文对您了解如何利用Redis存储复杂对象有所帮助。
参考资料