System-design-primer 学习如何设计大型系统为系统设计面试做准备

Short timeline – Aim for breadth with system design topics. Practice by solving some interview questions.
Medium timeline – Aim for breadth and some depth with system design topics. Practice by solving many interview questions.
Long timeline – Aim for breadth and more depth with system design topics. Practice by solving most interview questions.

	Short	Medium	Long
Read through the System design topics to get a broad understanding of how systems work	:+1:	:+1:	:+1:
Read through a few articles in the Company engineering blogs for the companies you are interviewing with	:+1:	:+1:	:+1:
Read through a few Real world architectures	:+1:	:+1:	:+1:
Review How to approach a system design interview question	:+1:	:+1:	:+1:
Work through System design interview questions with solutions	Some	Many	Most
Work through Object-oriented design interview questions with solutions	Some	Many	Most
Review Additional system design interview questions	Some	Many	Most

面向对象的设计面试问题及其解决方案

如何进行撞击系统设计面试题

系统设计面试是一场开放式的谈话。希望你来领导它

您可以使用以下步骤来指导讨论。要帮助巩固此过程，请使用以下步骤完成系统设计面试问题与解决方案部分

步骤3：设计核心组件

收集需求并确定问题范围。提出问题以澄清用例和约束。讨论假设

Who is going to use it?
How are they going to use it?
How many users are there?
What does the system do?
What are the inputs and outputs of the system?
How much data do we expect to handle?
How many requests per second do we expect?
What is the expected read to write ratio?

步骤4：调整设计比例

概述包含所有重要组件的高级设计

Sketch the main components and connections
Justify your ideas

粗略计算

深入了解每个核心组件的详细信息。例如，如果您被要求设计一个url缩短服务，请讨论：

Generating and storing a hash of the full url
- MD5 and Base62
- Hash collisions
- SQL or NoSQL
- Database schema
Translating a hashed url to the full url
- Database lookup
API and object-oriented design

来源和进一步阅读

在给定约束的情况下，确定并解决瓶颈问题。例如，您是否需要以下内容来解决可伸缩性问题？

Load balancer
Horizontal scaling
Caching
Database sharding

讨论潜在的解决方案和权衡。每件事都是权衡的。使用可扩展系统设计原则解决瓶颈问题

Design Pastebin.com(或bit.ly)

你可能会被要求手工做一些估算。有关以下资源，请参阅附录：

Use back of the envelope calculations
Powers of two table
Latency numbers every programmer should know

设计Twitter时间表和搜索(或Facebook提要和搜索)

请查看以下链接，以更好地了解预期内容：

系统设计主题：从此处开始

带有示例讨论、代码和图表的常见系统设计面试问题

链接到解决方案/文件夹中内容的解决方案

Question
Design Pastebin.com (or Bit.ly)	Solution
Design the Twitter timeline and search (or Facebook feed and search)	Solution
Design a web crawler	Solution
Design Mint.com	Solution
Design the data structures for a social network	Solution
Design a key-value store for a search engine	Solution
Design Amazon’s sales ranking by category feature	Solution
Design a system that scales to millions of users on AWS	Solution
Add a system design question	Contribute

设计一个网络爬虫程序

查看练习和解决方案

Design Mint.com

查看练习和解决方案

设计社交网络的数据结构

查看练习和解决方案

为搜索引擎设计键值存储

查看练习和解决方案

按类别功能设计亚马逊的销售排名

查看练习和解决方案

设计可扩展到数百万AWS用户的系统

查看练习和解决方案

第1步：复习可扩展性视频课程

查看练习和解决方案

步骤2：查看可伸缩性文章

查看练习和解决方案

性能与可扩展性

常见的面向对象设计面试问题，带有示例讨论、代码和图表

注：此部分正在开发中。

是系统设计的新手吗？

Question
Design a hash map	Solution
Design a least recently used cache	Solution
Design a call center	Solution
Design a deck of cards	Solution
Design a parking lot	Solution
Design a chat server	Solution
Design a circular array	Contribute
Add an object-oriented design question	Contribute

延迟与吞吐量

首先，您需要基本了解通用原则，了解它们是什么、如何使用以及它们的优缺点

哈佛大学可伸缩性讲座

下一步

可扩展性

Topics covered:
- Vertical scaling
- Horizontal scaling
- Caching
- Load balancing
- Database replication
- Database partitioning

盖子定理

接下来，我们来看看高级权衡：

Topics covered:
- Clones
- Databases
- Caches
- Asynchronism

弱一致性

请记住，每件事都是权衡的。

Performance vs scalability
Latency vs throughput
Availability vs consistency

然后，我们将深入探讨更具体的主题，如DNS、CDN和负载均衡器

如果服务以与添加的资源成正比的方式提高性能，则该服务是可伸缩的。通常，提高性能意味着服务更多的工作单元，但也可以处理更大的工作单元，例如当数据集增长时。1

可用性与一致性

看待性能与可伸缩性的另一种方式：

延迟是执行某种操作或产生某种结果的时间

If you have a performance problem, your system is slow for a single user.
If you have a scalability problem, your system is fast for a single user but slow under heavy load.

最终一致性

一致性模式

吞吐量是每单位时间内此类操作或结果的数量

通常，您应该以具有可接受延迟的最大吞吐量为目标

资料来源：复习上限定理

强一致性

Understanding latency vs throughput

可用性模式

故障转移

在分布式计算机系统中，您只能支持以下两项保证：

网络不可靠，因此您需要支持分区容错。您需要在一致性和可用性之间进行软件权衡

Consistency – Every read receives the most recent write or an error
Availability – Every request receives a response, without guarantee that it contains the most recent version of the information
Partition Tolerance – The system continues to operate despite arbitrary partitioning due to network failures

等待来自分区节点的响应可能会导致超时错误。如果您的业务需求需要原子读写，CP是一个很好的选择

SQL调优

响应返回任何节点上可用的最容易获得的数据版本，该版本可能不是最新版本。在解析分区时，写入可能需要一些时间才能传播

键值存储

如果业务需要考虑到最终的一致性，或者当系统需要在出现外部错误的情况下继续工作时，AP是一个很好的选择

对于同一数据的多个副本，我们面临着如何对它们进行同步操作的选择，以便客户对数据有一致的看法。回想一下CAP定理中的一致性定义-每次读取都会收到最近的写入或错误

缺点：故障转移

域名系统

写入后，读取可能会也可能看不到它。采取了尽力而为的方法

复制

这种方法可以在memcached等系统中看到。弱一致性适用于VoIP、视频聊天和实时多人游戏等实时用例。例如，如果您正在打电话，并且在几秒钟内失去接收，那么当您重新连接时，您听不到在连接中断期间所说的话

写入之后，读取最终会看到它(通常在毫秒内)。异步复制数据

数量上的可用性

在DNS和电子邮件等系统中可以看到这种方法。最终一致性在高可用性系统中运行良好

写入后，Reads将看到它。同步复制数据

缺点：DNS

这种方法可以在文件系统和RDBMS中看到。强一致性在需要事务的系统中运行良好

支持高可用性有两种互补模式：故障切换和复制

推流CDN

Transactions across data centers

内容交付网络

使用主动-被动故障转移时，会在处于备用状态的主动服务器和被动服务器之间发送心跳。如果心跳中断，被动服务器将接管主动服务器的IP地址并恢复服务

拉取CDN

文档存储

停机时间的长短取决于被动服务器是否已经在“热”待机状态下运行，或者它是否需要从“冷”待机状态启动。只有活动服务器才能处理流量

主动-被动故障切换也可以称为主-从故障切换

在主动-主动模式下，两台服务器都在管理流量，在它们之间分担负载

宽列存储

如果服务器是面向公众的，则DNS需要知道两个服务器的公共IP。如果服务器是面向内部的，则应用程序逻辑需要了解这两个服务器

主动-主动故障切换也可以称为主-主故障切换

此主题将在数据库部分进一步讨论：

劣势：CDN

Fail-over adds more hardware and additional complexity.
There is a potential for loss of data if the active system fails before any newly written data can be replicated to the passive.

第4层负载均衡

图形数据库

可用性通常通过正常运行时间(或停机时间)作为服务可用时间的百分比来量化。可用性通常用数字9来衡量–99.99%可用性的服务被描述为有四个9

Master-slave replication
Master-master replication

第7层负载均衡

如果服务由多个容易发生故障的组件组成，则服务的总体可用性取决于这些组件是顺序的还是并行的

来源和进一步阅读：NoSQL

Duration	Acceptable downtime
Downtime per year	8h 45min 57s
Downtime per month	43m 49.7s
Downtime per week	10m 4.8s
Downtime per day	1m 26.4s

后备缓存

Duration	Acceptable downtime
Downtime per year	52min 35.7s
Downtime per month	4m 23s
Downtime per week	1m 5s
Downtime per day	8.6s

直写

当两个可用性<100%的组件按顺序排列时，总体可用性会降低：

In sequence

如果Foo和Bar都有99.9%的可用性，那么它们的总可用性依次为99.8%

Availability (Total) = Availability (Foo) * Availability (Bar)

当两个可用性<100%的组件并行时，总体可用性会提高：

In parallel

如果Foo和BAR都有99.9%的可用性，那么它们并行的总可用性将是99.9999%

Availability (Total) = 1 - (1 - Availability (Foo)) * (1 - Availability (Bar))

来源：DNS安全演示

负载均衡器

域名系统(DNS)将域名(如www.example.com)转换为IP地址

DNS是分层的，顶层有几个权威服务器。您的路由器或ISP提供有关执行查找时联系哪些DNS服务器的信息。较低级别的DNS服务器缓存映射，这些映射可能会因DNS传播延迟而变得陈旧。您的浏览器或操作系统还可以将DNS结果缓存一段时间，具体取决于生存时间(TTL)

CloudFlare和Route 53等服务提供托管DNS服务。某些DNS服务可以通过各种方法路由流量：

NS record (name server) – Specifies the DNS servers for your domain/subdomain.
MX record (mail exchange) – Specifies the mail servers for accepting messages.
A record (address) – Points a name to an IP address.
CNAME (canonical) – Points a name to another name or CNAME (example.com to www.example.com) or to an A record.

来源：为什么使用CDN

Weighted round robin
- Prevent traffic from going to servers under maintenance
- Balance between varying cluster sizes
- A/B testing
Latency-based
Geolocation-based

水平缩放

Accessing a DNS server introduces a slight delay, although mitigated by caching described above.
DNS server management could be complex and is generally managed by governments, ISPs, and large companies.
DNS services have recently come under DDoS attack, preventing users from accessing websites such as Twitter without knowing Twitter’s IP address(es).

缺点：负载均衡器

反向代理(Web服务器)

内容递送网络(CDN)是全球分布的代理服务器网络，提供来自更靠近用户的位置的内容。一般情况下，静电文件(如html/css/js)、照片和视频都是由云服务提供的，但有些云服务(如亚马逊的云前端)支持动态内容。站点的DNS解析将告诉客户端要联系哪个服务器

从CDN提供内容可以通过两种方式显著提高性能：

每当您的服务器发生更改时，推送CDN都会接收新内容。您负责提供内容、直接上传到CDN、重写指向CDN的URL。您可以配置内容何时过期以及何时更新。只有在内容是新的或更改的情况下才会上载内容，从而最大限度地减少流量，但最大限度地提高存储

Users receive content from data centers close to them
Your servers do not have to serve requests that the CDN fulfills

负载均衡器与反向代理

流量较小的站点或内容不经常更新的站点可以很好地使用推送CDN。内容只放在CDN上一次，而不是定期重新拉取

拉取CDN在第一个用户请求内容时从您的服务器抓取新内容。您将内容保留在服务器上，并重写URL以指向CDN。这会导致请求速度变慢，直到内容缓存在CDN上

缺点：反向代理

生存时间(TTL)确定缓存内容的时间长度。拉取CDN最大限度地减少了CDN上的存储空间，但如果文件过期并在实际更改之前被拉取，则可能会产生冗余流量

流量大的站点可以很好地使用拉式CDN，因为流量分布更均匀，只有最近请求的内容保留在CDN上

来源：可伸缩系统设计模式

微服务

CDN costs could be significant depending on traffic, although this should be weighed with additional costs you would incur not using a CDN.
Content might be stale if it is updated before the TTL expires it.
CDNs require changing URLs for static content to point to the CDN.

服务发现

应用层

负载平衡器将传入的客户端请求分发到应用程序服务器和数据库等计算资源。在每种情况下，负载均衡器都会将来自计算资源的响应返回到相应的客户端。负载均衡器在以下方面有效：

负载均衡器可以通过硬件(昂贵)或软件(如HAProxy)实现

Preventing requests from going to unhealthy servers
Preventing overloading resources
Helping to eliminate a single point of failure

其他优势包括：

为了防止故障，通常在主动-被动或主动-主动模式下设置多个负载均衡器

SSL termination – Decrypt incoming requests and encrypt server responses so backend servers do not have to perform these potentially expensive operations
- Removes the need to install X.509 certificates on each server
Session persistence – Issue cookies and route a specific client’s requests to same instance if the web apps do not keep track of sessions

负载均衡器可以根据各种指标路由流量，包括：

第4层负载均衡器查看传输层的信息，以决定如何分发请求。通常，这涉及报头中的源IP地址、目的IP地址和端口，但不涉及数据包的内容。第4层负载均衡器转发进出上游服务器的网络数据包，执行网络地址转换(NAT)

Random
Least loaded
Session/cookies
Round robin or weighted round robin
Layer 4
Layer 7

缺点：应用层

第7层负载均衡器查看应用层以决定如何分发请求。这可能涉及标头、消息和Cookie的内容。第7层负载均衡器终止网络流量，读取消息，做出负载平衡决策，然后打开到所选服务器的连接。例如，第7层负载均衡器可以将视频流量定向到托管视频的服务器，同时将更敏感的用户计费流量定向到经过安全强化的服务器

关系数据库管理系统(RDBMS)

以灵活性为代价，与第7层相比，第4层负载平衡需要更少的时间和计算资源，尽管对现代商用硬件的性能影响可能微乎其微

负载均衡器还可以帮助进行水平扩展，从而提高性能和可用性。与在更昂贵的硬件上纵向扩展单个服务器(称为垂直扩展)相比，使用商用计算机进行横向扩展更具成本效益，并带来更高的可用性。与专门的企业系统相比，在商用硬件上工作的人才也更容易招聘

NoSQL

来源：维基百科

写后(回写)

Scaling horizontally introduces complexity and involves cloning servers
- Servers should be stateless: they should not contain any user-related data like sessions or profile pictures
- Sessions can be stored in a centralized data store such as a database (SQL, NoSQL) or a persistent cache (Redis, Memcached)
Downstream servers such as caches and databases need to handle more simultaneous connections as upstream servers scale out

SQL或NoSQL

The load balancer can become a performance bottleneck if it does not have enough resources or if it is not configured properly.
Introducing a load balancer to help eliminate a single point of failure results in increased complexity.
A single load balancer is a single point of failure, configuring multiple load balancers further increases complexity.

客户端缓存

数据库

反向代理是集中内部服务并向公众提供统一接口的Web服务器。在反向代理将服务器的响应返回给客户端之前，将来自客户端的请求转发到可以实现该请求的服务器

来源：规模架构系统简介

通过将Web层与应用层(也称为平台层)分开，您可以分别扩展和配置这两个层。添加新API会导致添加应用程序服务器，而不必添加额外的Web服务器。单一责任原则主张共同工作的小型自主服务。拥有小型服务的小型团队可以更积极地规划快速增长

Increased security – Hide information about backend servers, blacklist IPs, limit number of connections per client
Increased scalability and flexibility – Clients only see the reverse proxy’s IP, allowing you to scale servers or change their configuration
SSL termination – Decrypt incoming requests and encrypt server responses so backend servers do not have to perform these potentially expensive operations
- Removes the need to install X.509 certificates on each server
Compression – Compress server responses
Caching – Return the response for cached requests
Static content – Serve static content directly
- HTML/CSS/JS
- Photos
- Videos
- Etc

CDN缓存

Deploying a load balancer is useful when you have multiple servers. Often, load balancers route traffic to a set of servers serving the same function.
Reverse proxies can be useful even with just one web server or application server, opening up the benefits described in the previous section.
Solutions such as NGINX and HAProxy can support both layer 7 reverse proxying and load balancing.

Web服务器缓存

Introducing a reverse proxy results in increased complexity.
A single reverse proxy is a single point of failure, configuring multiple reverse proxies (ie a failover) further increases complexity.

数据库缓存

高速缓存

应用程序层中的工作者也有助于启用异步

与此讨论相关的是微服务，可以将其描述为一套可独立部署的小型模块化服务。每个服务都运行一个独特的流程，并通过定义良好的轻量级机制进行通信，以服务于业务目标。1个

例如，Pinterest可以拥有以下微服务：用户档案、追随者、馈送、搜索、照片上传等

应用程序缓存

诸如Consul、etcd和ZooKeeper这样的系统可以通过跟踪注册的名称、地址和端口来帮助服务找到彼此。运行状况检查有助于验证服务完整性，通常使用HTTP端点来完成。Consul和etcd都有一个内置的键值存储，可用于存储配置值和其他共享数据

来源：向上扩展至您的第一个1000万用户

数据库查询级别的高速缓存

像SQL这样的关系数据库是以表形式组织的数据项的集合

对象级别的缓存

Adding an application layer with loosely coupled services requires a different approach from an architectural, operations, and process viewpoint (vs a monolithic system).
Microservices can add complexity in terms of deployments and operations.

何时更新缓存

异步

ACID是关系数据库事务的一组属性

缺点：缓存

扩展关系数据库有许多技术：主-从复制、主-主复制、联合、分片、反规范化和SQL调优

主机服务于读取和写入，将写入复制到一个或多个仅服务于读取的从机。从属设备还可以以树状方式复制到其他从属设备。如果主机脱机，系统可以继续以只读模式运行，直到将从属提升为主机或调配新主机

Atomicity – Each transaction is all or nothing
Consistency – Any transaction will bring the database from one valid state to another
Isolation – Executing transactions concurrently has the same results as if the transactions were executed serially
Durability – Once a transaction has been committed, it will remain so

来源：可伸缩性、可用性、稳定性、模式

提前刷新

两个主机都提供读写服务，并在写入时相互协调。如果任一主机发生故障，系统可以在读取和写入的情况下继续运行

来源：可伸缩性、可用性、稳定性、模式

Disadvantage(s): master-slave replication

Additional logic is needed to promote a slave to a master.
See Disadvantage(s): replication for points related to both master-slave and master-master.

来源和进一步阅读：HTTP

来源：向上扩展至您的第一个1000万用户

联合(或功能分区)按功能拆分数据库。例如，您可以拥有三个数据库，而不是一个单一的整体数据库：论坛、用户和产品，从而减少每个数据库的读写流量，从而减少复制延迟。较小的数据库会产生更多可以放入内存的数据，这反过来又会因为改进的高速缓存位置而导致更多的高速缓存命中率。由于没有单个中央主机串行化写入，您可以并行写入，从而增加吞吐量

Disadvantage(s): master-master replication

You’ll need a load balancer or you’ll need to make changes to your application logic to determine where to write.
Most master-master systems are either loosely consistent (violating ACID) or have increased write latency due to synchronization.
Conflict resolution comes more into play as more write nodes are added and as latency increases.
See Disadvantage(s): replication for points related to both master-slave and master-master.

Disadvantage(s): replication

There is a potential for loss of data if the master fails before any newly written data can be replicated to other nodes.
Writes are replayed to the read replicas. If there are a lot of writes, the read replicas can get bogged down with replaying writes and can’t do as many reads.
The more read slaves, the more you have to replicate, which leads to greater replication lag.
On some systems, writing to the master can spawn multiple threads to write in parallel, whereas read replicas only support writing sequentially with a single thread.
Replication adds more hardware and additional complexity.

Source(s) and further reading: replication

来源和进一步阅读：TCP和UDP

来源：可伸缩性、可用性、稳定性、模式

分片将数据分布在不同的数据库中，以便每个数据库只能管理数据的一个子集。以用户数据库为例，随着用户数量的增加，集群中会添加更多的分片

Disadvantage(s): federation

Federation is not effective if your schema requires huge functions or tables.
You’ll need to update your application logic to determine which database to read and write.
Joining data from two databases is more complex with a server link.
Federation adds more hardware and additional complexity.

Source(s) and further reading: federation

Scaling up to your first 10 million users

缺点：RPC

与联合的优点类似，分片可以减少读写流量、减少复制和增加缓存命中率。索引大小也会减小，这通常会通过更快的查询提高性能。如果一个碎片发生故障，其他碎片仍可运行，尽管您需要添加某种形式的复制以避免数据丢失。与联合一样，没有单个串行化写入的中央主机，允许您在提高吞吐量的同时并行写入

共享用户表的常见方式是通过用户的姓氏首字母或用户的地理位置

反规格化试图以牺牲一些写入性能为代价来提高读取性能。数据的冗余副本被写入多个表中，以避免昂贵的联接。一些RDBMS(如PostgreSQL和Oracle)支持物化视图，物化视图处理存储冗余信息和保持冗余副本一致的工作

使用联合和分片等技术分发数据后，管理跨数据中心的联接会进一步增加复杂性。反规格化可能会绕过对这种复杂连接的需要。

Disadvantage(s): sharding

You’ll need to update your application logic to work with shards, which could result in complex SQL queries.
Data distribution can become lopsided in a shard. For example, a set of power users on a shard could result in increased load to that shard compared to others.
- Rebalancing adds additional complexity. A sharding function based on consistent hashing can reduce the amount of transferred data.
Joining data from multiple shards is more complex.
Sharding adds more hardware and additional complexity.

Source(s) and further reading: sharding

劣势：睡觉

在大多数系统中，读取的数量可能远远超过写入的数量100：1甚至1000：1。导致复杂数据库联接的读取可能非常昂贵，会花费大量时间进行磁盘操作

SQL调优是一个涉及面很广的主题，很多书都是作为参考编写的

重要的是要进行基准测试和性能分析，以模拟和发现瓶颈

Disadvantage(s): denormalization

Data is duplicated.
Constraints can help redundant copies of information stay in sync, which increases complexity of the database design.
A denormalized database under heavy write load might perform worse than its normalized counterpart.

Source(s) and further reading: denormalization

Denormalization

来源及进一步阅读：睡觉和rpc

基准测试和性能分析可能会为您提供以下优化

NoSQL是键值存储、文档存储、宽列存储或图形数据库中表示的数据项的集合。数据被反规范化，连接通常在应用程序代码中完成。大多数NoSQL存储缺乏真正的ACID事务，倾向于最终的一致性

Benchmark – Simulate high-load situations with tools such as ab.
Profile – Enable tools such as the slow query log to help track performance issues.

BASE通常用来描述NoSQL数据库的属性。与CAP定理相比，BASE选择可用性而不是一致性

Tighten up the schema

MySQL dumps to disk in contiguous blocks for fast access.
Use CHAR instead of VARCHAR for fixed-length fields.
- CHAR effectively allows for fast, random access, whereas with VARCHAR, you must find the end of a string before moving onto the next one.
Use TEXT for large blocks of text such as blog posts. TEXT also allows for boolean searches. Using a TEXT field results in storing a pointer on disk that is used to locate the text block.
Use INT for larger numbers up to 2^32 or 4 billion.
Use DECIMAL for currency to avoid floating point representation errors.
Avoid storing large BLOBS, store the location of where to get the object instead.
VARCHAR(255) is the largest number of characters that can be counted in an 8 bit number, often maximizing the use of a byte in some RDBMS.
Set the NOT NULL constraint where applicable to improve search performance.

Use good indices

Columns that you are querying (SELECT, GROUP BY, ORDER BY, JOIN) could be faster with indices.
Indices are usually represented as self-balancing B-tree that keeps data sorted and allows searches, sequential access, insertions, and deletions in logarithmic time.
Placing an index can keep the data in memory, requiring more space.
Writes could also be slower since the index also needs to be updated.
When loading large amounts of data, it might be faster to disable indices, load the data, then rebuild the indices.

Avoid expensive joins

Denormalize where performance demands it.

Partition tables

Break up a table by putting hot spots in a separate table to help keep it in memory.

Tune the query cache

In some cases, the query cache could lead to performance issues.

Source(s) and further reading: SQL tuning

消息队列

除了在SQL或NoSQL之间进行选择之外，了解哪种类型的NoSQL数据库最适合您的用例也很有帮助。在下一节中，我们将回顾键值存储、文档存储、宽列存储和图形数据库

抽象：哈希表

Basically available – the system guarantees availability.
Soft state – the state of the system may change over time, even without input.
Eventual consistency – the system will become consistent over a period of time, given that the system doesn’t receive input during that period.

键值存储通常允许O(1)次读取和写入，并且通常由内存或SSD支持。数据存储可以按字典顺序维护键，从而允许高效地检索键范围。键值存储可允许存储具有值的元数据

来源和进一步阅读

键值存储提供高性能，通常用于简单数据模型或快速变化的数据，如内存缓存层。由于它们只提供有限的操作集，因此如果需要额外的操作，复杂性将转移到应用层

键值存储是更复杂系统(如文档存储，在某些情况下还包括图形数据库)的基础

抽象：键值存储，文档存储为值

文档存储以文档(XML、JSON、二进制等)为中心，文档存储给定对象的所有信息。文档存储提供基于文档本身的内部结构进行查询的API或查询语言。请注意，许多键值存储包括用于使用值的元数据的功能，从而模糊了这两种存储类型之间的界限

Source(s) and further reading: key-value store

可视化的延迟数字

根据底层实现，文档按集合、标签、元数据或目录进行组织。尽管可以将文档组织或分组在一起，但文档的字段可能彼此完全不同

一些文档存储，如MongoDB和CouchDB，也提供了一种类似SQL的语言来执行复杂的查询。DynamoDB同时支持键值和文档

文档存储具有很高的灵活性，通常用于处理偶尔更改的数据

来源：SQL&NoSQL，简史

抽象：嵌套映射ColumnFamily<RowKey，Columns<ColKey，Value，Timestamp>>

Source(s) and further reading: document store

英语∙日本語∙简体中文∙繁體中文|العَرَبِيَّة‎∙বাংলা∙葡萄牙语do Brasil∙Deutsch∙ελληνικά∙עברית∙Italiano∙한국어∙فارسی∙Poliano∙русскийязык∙Español∙ภาษาไทย∙Türkçe∙tiếng Việt∙Français|添加翻译

宽列存储的基本数据单位是列(名称/值对)。列可以按列族分组(类似于SQL表)。超级柱族进一步将柱族分组。您可以使用行键单独访问每列，具有相同行键的列形成一行。每个值都包含一个用于版本化和冲突解决的时间戳

Google引入了Bigtable作为第一个宽列存储，它影响了Hadoop生态系统中经常使用的开源HBase，以及Facebook的Cassandra。Bigtable、HBase和Cassandra等存储按字典顺序维护键，从而允许高效地检索选择性键范围

宽列存储提供高可用性和高可伸缩性。它们通常用于非常大的数据集

来源：图表数据库

抽象：图表

Source(s) and further reading: wide column store

网络不可靠，因此您需要支持分区容错。您需要在一致性和可用性之间进行软件权衡

在图形数据库中，每个节点是一条记录，每条弧是两个节点之间的关系。对图形数据库进行了优化，以表示具有多个外键或多对多关系的复杂关系

图形数据库为具有复杂关系的数据模型(如社交网络)提供高性能。它们相对较新，尚未广泛使用；可能更难找到开发工具和资源。很多图表只能通过睡觉API访问

来源：从RDBMS过渡到NoSQL

SQL的原因：

Source(s) and further reading: graph

请注意，许多键值存储包括用于使用值的元数据的功能，从而模糊了这两种存储类型之间的界限

任务队列

使用NoSQL的原因：

非常适合NoSQL的示例数据：

Structured data
Strict schema
Relational data
Need for complex joins
Transactions
Clear patterns for scaling
More established: developers, community, code, tools, etc
Lookups by index are very fast

来源：可伸缩系统设计模式

Semi-structured data
Dynamic or flexible schema
Non-relational data
No need for complex joins
Store many TB (or PB) of data
Very data intensive workload
Very high throughput for IOPS

缓存可以缩短页面加载时间，并可以减少服务器和数据库的负载。在此模型中，调度程序将首先查找以前是否已发出请求，并尝试查找要返回的前一个结果，以便保存实际执行

Rapid ingest of clickstream and log data
Leaderboard or scoring data
Temporary data, such as a shopping cart
Frequently accessed (‘hot’) tables
Metadata/lookup tables

Source(s) and further reading: SQL or NoSQL

沟通

数据库通常受益于跨其分区的读和写的统一分布。受欢迎的项目可能会扭曲分布，造成瓶颈。将缓存放在数据库前面有助于吸收不均匀的负载和流量高峰

缓存可以位于客户端(操作系统或浏览器)、服务器端或位于不同的缓存层中

CDN被认为是一种缓存

背压

反向代理和缓存(如varish)可以直接服务于静电和动态内容。Web服务器还可以缓存请求，无需联系应用程序服务器即可返回响应

缺点：异步

您的数据库通常在默认配置中包含某种级别的缓存，针对一般用例进行了优化。针对特定的使用模式调整这些设置可以进一步提高性能

超文本传输协议(HTTP)

内存缓存(如memcached和redis)是应用程序和数据存储之间的键值存储。由于数据保存在RAM中，因此它比数据存储在磁盘上的典型数据库快得多。RAM比磁盘更有限，因此缓存失效算法(如最近最少使用(LRU))可以帮助使“冷”条目无效，并将“热”数据保留在RAM中

传输控制协议(TCP)

Redis具有以下附加功能：

用户数据报协议(UDP)

您可以缓存多个级别，分为两个一般类别：数据库查询和对象：

通常，您应该尽量避免基于文件的缓存，因为这会增加克隆和自动缩放的难度

Persistence option
Built-in data structures such as sorted sets and lists

无论何时查询数据库，都要将查询作为键进行散列，并将结果存储到缓存中。此方法存在过期问题：

Row level
Query-level
Fully-formed serializable objects
Fully-rendered HTML

将数据视为对象，类似于您对应用程序代码所做的操作。让您的应用程序将数据库中的数据集组装成一个类实例或一个或多个数据结构：

远程过程调用(RPC)

缓存内容的建议：

Hard to delete a cached result with complex queries
If one piece of data changes such as a table cell, you need to delete all cached queries that might include the changed cell

表述性状态转移(睡觉)

由于您只能在缓存中存储有限数量的数据，因此您需要确定哪种缓存更新策略最适合您的用例

Remove the object from cache if its underlying data has changed
Allows for asynchronous processing: workers assemble objects by consuming the latest cached object

来源：从缓存到内存中数据网格

User sessions
Fully rendered web pages
Activity streams
User graph data

远程过程调用与睡觉调用比较

应用程序负责从存储中读取和写入。缓存不直接与存储交互。应用程序执行以下操作：

我在开放源码许可下向您提供此存储库中的代码和资源。因为这是我的个人存储库，您获得的我的代码和资源的许可证来自我，而不是我的雇主(Facebook)

memcached通常以这种方式使用

后续读取添加到高速缓存的数据速度很快。侧缓存也称为惰性加载。仅缓存请求的数据，从而避免使用未请求的数据填满缓存

Look for entry in cache, resulting in a cache miss
Load entry from the database
Add entry to cache
Return entry

def get_user(self, user_id):
    user = cache.get("user.{0}", user_id)
    if user is None:
        user = db.query("SELECT * FROM users WHERE user_id = {0}", user_id)
        if user is not None:
            key = "user.{0}".format(user_id)
            cache.set(key, json.dumps(user))
    return user

来源：可伸缩性、可用性、稳定性、模式

应用程序使用高速缓存作为主数据存储，对其进行读写数据，而高速缓存负责对数据库进行读写：

Disadvantage(s): cache-aside

Each cache miss results in three trips, which can cause a noticeable delay.
Data can become stale if it is updated in the database. This issue is mitigated by setting a time-to-live (TTL) which forces an update of the cache entry, or by using write-through.
When a node fails, it is replaced by a new, empty node, increasing latency.

Write-through

应用程序代码：

缓存代码：

Application adds/updates entry in cache
Cache synchronously writes entry to data store
Return

由于写入操作，直写操作的总体速度较慢，但后续读取刚写入的数据会很快。用户在更新数据时通常比读取数据时更能容忍延迟。缓存中的数据未过时

set_user(12345, {"foo":"bar"})

来源：可伸缩性、可用性、稳定性、模式

def set_user(user_id, values):
    user = db.query("UPDATE Users WHERE id = {0}", user_id, values)
    cache.set(user_id, user)

在Write-Back中，应用程序执行以下操作：

Disadvantage(s): write through

When a new node is created due to failure or scaling, the new node will not cache entries until the entry is updated in the database. Cache-aside in conjunction with write through can mitigate this issue.
Most data written might never be read, which can be minimized with a TTL.

Write-behind (write-back)

来源：从缓存到内存中数据网格

您可以将缓存配置为在任何最近访问的缓存条目到期之前自动刷新

Add/update entry in cache
Asynchronously write entry to the data store, improving write performance

Disadvantage(s): write-behind

There could be data loss if the cache goes down prior to its contents hitting the data store.
It is more complex to implement write-behind than it is to implement cache-aside or write-through.

Refresh-ahead

如果缓存可以准确预测将来可能需要哪些项目，则与直读相比，提前刷新可以降低延迟

来源：规模架构系统简介

异步工作流有助于减少代价高昂的操作的请求时间，否则这些操作将以内联方式执行。他们还可以通过提前执行耗时的工作来提供帮助，例如定期聚合数据

Disadvantage(s): refresh-ahead

Not accurately predicting which items are likely to be needed in the future can result in reduced performance than without refresh-ahead.

二次幂表

Need to maintain consistency between caches and the source of truth such as the database through cache invalidation.
Cache invalidation is a difficult problem, there is additional complexity associated with when to update the cache.
Need to make application changes such as adding Redis or memcached.

每个程序员都应该知道的延迟数字

安全性

消息队列接收、保存和传递消息。如果操作速度太慢而无法以内联方式执行，则可以将消息队列与以下工作流结合使用：

用户不会被阻止，作业将在后台处理。在此期间，客户端可能会选择性地进行少量处理，使其看起来好像任务已经完成。例如，如果发布一条tweet，该tweet可以立即发布到您的时间表上，但可能需要一段时间才能真正将您的tweet发送给您的所有追随者

其他系统设计面试问题

Redis作为简单的消息代理很有用，但消息可能会丢失

An application publishes a job to the queue, then notifies the user of job status
A worker picks up the job from the queue, processes it, then signals the job is complete

RabbitMQ很流行，但需要您适应‘AMQP’协议并管理您自己的节点

Amazon SQS是托管的，但可能会有很高的延迟，并且可能会传递两次消息

任务队列接收任务及其相关数据，运行它们，然后交付结果。它们可以支持调度，并可用于在后台运行计算密集型作业

芹菜支持调度，并且主要支持python。

现实世界的建筑

如果队列开始显著增长，队列大小可能会大于内存，从而导致缓存未命中、磁盘读取，甚至会降低性能。反压可以通过限制队列大小来提供帮助，从而为队列中已有的作业保持较高的吞吐率和良好的响应时间。一旦队列填满，客户端就会收到服务器繁忙或HTTP 503状态代码，以便稍后重试。客户端可以稍后重试请求，可能会使用指数回退

来源：OSI 7层模型

公司架构

HTTP是一种在客户端和服务器之间编码和传输数据的方法。它是一种请求/响应协议：客户端发出请求，服务器发出响应，其中包含请求的相关内容和完成状态信息。HTTP是独立的，允许请求和响应流经许多执行负载平衡、缓存、加密和压缩的中间路由器和服务器

公司工程博客

Use cases such as inexpensive calculations and realtime workflows might be better suited for synchronous operations, as introducing queues can add delays and complexity.

CP-一致性和划分容错

附录

基本HTTP请求由谓词(方法)和资源(端点)组成。以下是常见的HTTP谓词：

AP-可用性和分区容错

*可以多次调用，没有不同的结果

HTTP是依赖于较低级别协议(如TCP和UDP)的应用层协议

Verb	Description	Idempotent*	Safe	Cacheable
GET	Reads a resource	Yes	Yes	Yes
POST	Creates a resource or trigger a process that handles data	No	No	Yes if response contains freshness info
PUT	Creates or replace a resource	Yes	No	No
PATCH	Partially updates a resource	No	No	Yes if response contains freshness info
DELETE	Deletes a resource	Yes	No	No

来源：如何制作多人游戏

TCP是IP网络上的面向连接的协议。使用握手建立和终止连接。所有发送的数据包都保证按原始顺序到达目的地，并且通过以下方式不会损坏：

Source(s) and further reading: HTTP

主动-被动

如果发送方没有收到正确的响应，它将重新发送数据包。如果存在多个超时，则连接将断开。TCP还实施流量控制和拥塞控制。这些保证会导致延迟，并且通常会导致传输效率低于UDP

为了确保高吞吐量，Web服务器可以保持大量TCP连接处于打开状态，从而导致高内存使用率。在web服务器线程和memcached服务器(比方说)之间具有大量打开的连接可能代价高昂。除了在适用的情况下切换到UDP之外，连接池还可以提供帮助

Sequence numbers and checksum fields for each packet
Acknowledgement packets and automatic retransmission

TCP对于要求高可靠性但对时间要求较低的应用程序很有用。一些示例包括Web服务器、数据库信息、SMTP、FTP和SSH

在以下情况下使用UDP上的TCP：

来源：如何制作多人游戏

UDP是无连接的。数据报(类似于数据包)仅在数据报级别得到保证。数据报可能无序到达目的地，也可能根本没有到达目的地。UDP不支持拥塞控制。如果没有TCP支持的保证，UDP通常更有效

You need all of the data to arrive intact
You want to automatically make a best estimate use of the network throughput

主动-主动

UDP可以广播，向子网中的所有设备发送数据报。这对于DHCP很有用，因为客户端尚未接收到IP地址，因此阻止了TCP在没有IP地址的情况下流式传输

UDP的可靠性较低，但在VoIP、视频聊天、流和实时多人游戏等实时使用案例中运行良好

在以下情况下使用TCP上的UDP：

来源：破解系统设计访谈

在RPC中，客户端导致过程在不同的地址空间(通常是远程服务器)上执行。该过程的编码就好像它是一个本地过程调用，从客户端程序抽象出如何与服务器通信的细节。远程调用通常比本地调用更慢、更不可靠，因此区分RPC调用和本地调用很有帮助。流行的RPC框架包括Protobuf、Thrift和Avro

You need the lowest latency
Late data is worse than loss of data
You want to implement your own error correction

Source(s) and further reading: TCP and UDP

主-从和主-主

RPC是一种请求-响应协议：

示例RPC调用：

RPC专注于公开行为。RPC通常用于内部通信的性能原因，因为您可以手工创建本地调用以更好地适应您的用例

Client program – Calls the client stub procedure. The parameters are pushed onto the stack like a local procedure call.
Client stub procedure – Marshals (packs) procedure id and arguments into a request message.
Client communication module – OS sends the message from the client to the server.
Server communication module – OS passes the incoming packets to the server stub procedure.
Server stub procedure – Unmarshalls the results, calls the server procedure matching the procedure id and passes the given arguments.
The server response repeats the steps above in reverse order.

在以下情况下选择本机库(也称为SDK)：

GET /someoperation?data=anId

POST /anotheroperation
{
  "data":"anId";
  "anotherdata": "another value"
}

睡觉之后的HTTP接口倾向于更常用于公共接口

睡觉是一种实施客户端/服务器模型的体系结构风格，其中客户端作用于由服务器管理的一组资源。服务器提供资源和动作的表示，这些资源和动作既可以操作资源，也可以获得新的资源表示。所有通信必须是无状态和可缓存的

You know your target platform.
You want to control how your “logic” is accessed.
You want to control how error control happens off your library.
Performance and end user experience is your primary concern.

REST风格的界面有四个特点：

Disadvantage(s): RPC

RPC clients become tightly coupled to the service implementation.
A new API must be defined for every new operation or use case.
It can be difficult to debug RPC.
You might not be able to leverage existing technologies out of the box. For example, it might require additional effort to ensure RPC calls are properly cached on caching servers such as Squid.

99.9%可用性-三个9

睡觉调用示例：

睡觉专注于数据曝光。它最大限度地减少了客户端/服务器之间的耦合，通常用于公共HTTPAPI。睡觉使用一种更通用、更统一的方法，通过URI公开资源，通过头部表示，通过GET、POST、PUT、DELETE和PATCH等动词进行操作。由于是无状态的，睡觉非常适合水平伸缩和分区

Identify resources (URI in HTTP) – use the same URI regardless of any operation.
Change with representations (Verbs in HTTP) – use verbs, headers, and body.
Self-descriptive error message (status response in HTTP) – Use status codes, don’t reinvent the wheel.
HATEOAS (HTML interface for HTTP) – your web service should be fully accessible in a browser.

来源：你真的知道为什么你更喜欢睡觉而不是rpc吗？

GET /someresources/anId

PUT /someresources/anId
{"anotherdata": "another value"}

此部分可能需要一些更新。考虑捐款吧！

Disadvantage(s): REST

With REST being focused on exposing data, it might not be a good fit if resources are not naturally organized or accessed in a simple hierarchy. For example, returning all updated records from the past hour matching a particular set of events is not easily expressed as a path. With REST, it is likely to be implemented with a combination of URI path, query parameters, and possibly the request body.
REST typically relies on a few verbs (GET, POST, PUT, DELETE, and PATCH) which sometimes doesn’t fit your use case. For example, moving expired documents to the archive folder might not cleanly fit within these verbs.
Fetching complicated resources with nested hierarchies requires multiple round trips between the client and server to render single views, e.g. fetching content of a blog entry and the comments on that entry. For mobile applications operating in variable network conditions, these multiple roundtrips are highly undesirable.
Over time, more fields might be added to an API response and older clients will receive all new data fields, even those that they do not need, as a result, it bloats the payload size and leads to larger latencies.

99.99%可用性-四个9

Operation	RPC	REST
Signup	POST /signup	POST /persons
Resign	POST /resign { “personid”: “1234” }	DELETE /persons/1234
Read a person	GET /readPerson?personid=1234	GET /persons/1234
Read a person’s items list	GET /readUsersItemsList?personid=1234	GET /persons/1234/items
Add an item to a person’s items	POST /addItemToUsersItemsList { “personid”: “1234”; “itemid”: “456” }	POST /persons/1234/items { “itemid”: “456” }
Update an item	POST /modifyItem { “itemid”: “456”; “key”: “value” }	PUT /items/456 { “key”: “value” }
Delete an item	POST /removeItem { “itemid”: “456” }	DELETE /items/456

安全是一个广泛的话题。除非你有相当多的经验，有安全背景，或者正在申请一个需要安全知识的职位，否则你可能不需要知道更多的基础知识：

Source(s) and further reading: REST and RPC

正在开发中

有时你会被要求做“粗略”的估算。例如，您可能需要确定从磁盘生成100个图像缩略图需要多长时间，或者一个数据结构需要多少内存。每个程序员都应该知道的两个表的幂和延迟数字是很方便的参考资料

基于以上数字的便捷指标：

Encrypt in transit and at rest.
Sanitize all user inputs or any input parameters exposed to user to prevent XSS and SQL injection.
Use parameterized queries to prevent SQL injection.
Use the principle of least privilege.

并行可用性与顺序可用性

学分

缺点：水平缩放

Power           Exact Value         Approx Value        Bytes
---------------------------------------------------------------
7                             128
8                             256
10                           1024   1 thousand           1 KB
16                         65,536                       64 KB
20                      1,048,576   1 million            1 MB
30                  1,073,741,824   1 billion            1 GB
32                  4,294,967,296                        4 GB
40              1,099,511,627,776   1 trillion           1 TB

Source(s) and further reading

Powers of two

主从复制

Latency Comparison Numbers
--------------------------
L1 cache reference                           0.5 ns
Branch mispredict                            5   ns
L2 cache reference                           7   ns                      14x L1 cache
Mutex lock/unlock                           25   ns
Main memory reference                      100   ns                      20x L2 cache, 200x L1 cache
Compress 1K bytes with Zippy            10,000   ns       10 us
Send 1 KB bytes over 1 Gbps network     10,000   ns       10 us
Read 4 KB randomly from SSD*           150,000   ns      150 us          ~1GB/sec SSD
Read 1 MB sequentially from memory     250,000   ns      250 us
Round trip within same datacenter      500,000   ns      500 us
Read 1 MB sequentially from SSD*     1,000,000   ns    1,000 us    1 ms  ~1GB/sec SSD, 4X memory
HDD seek                            10,000,000   ns   10,000 us   10 ms  20x datacenter roundtrip
Read 1 MB sequentially from 1 Gbps  10,000,000   ns   10,000 us   10 ms  40x memory, 10X SSD
Read 1 MB sequentially from HDD     30,000,000   ns   30,000 us   30 ms 120x memory, 30X SSD
Send packet CA->Netherlands->CA    150,000,000   ns  150,000 us  150 ms

Notes
-----
1 ns = 10^-9 seconds
1 us = 10^-6 seconds = 1,000 ns
1 ms = 10^-3 seconds = 1,000 us = 1,000,000 ns

常见的系统设计面试问题，以及有关如何解决每个问题的资源链接

Read sequentially from HDD at 30 MB/s
Read sequentially from 1 Gbps Ethernet at 100 MB/s
Read sequentially from SSD at 1 GB/s
Read sequentially from main memory at 4 GB/s
6-7 world-wide round trips per second
2,000 round trips per second within a data center

Latency numbers visualized

有关如何设计真实系统的文章

Source(s) and further reading

主-主复制

来源：Twitter时间表按规模

Question	Reference(s)
Design a file sync service like Dropbox	youtube.com
Design a search engine like Google	queue.acm.org stackexchange.com ardendertat.com stanford.edu
Design a scalable web crawler like Google	quora.com
Design Google docs	code.google.com neil.fraser.name
Design a key-value store like Redis	slideshare.net
Design a cache system like Memcached	slideshare.net
Design a recommendation system like Amazon’s	hulu.com ijcai13.org
Design a tinyurl system like Bitly	n00tc0d3r.blogspot.com
Design a chat app like WhatsApp	highscalability.com
Design a picture sharing system like Instagram	highscalability.com highscalability.com
Design the Facebook news feed function	quora.com quora.com slideshare.net
Design the Facebook timeline function	facebook.com highscalability.com
Design the Facebook chat function	erlang-factory.com facebook.com
Design a graph search function like Facebook’s	facebook.com facebook.com facebook.com
Design a content delivery network like CloudFlare	figshare.com
Design a trending topic system like Twitter’s	michael-noll.com snikolov .wordpress.com
Design a random ID generation system	blog.twitter.com github.com
Return the top k requests during a time interval	cs.ucsb.edu wpi.edu
Design a system that serves data from multiple data centers	highscalability.com
Design an online multiplayer card game	indieflashblog.com buildnewgames.com
Design a garbage collection system	stuffwithstuff.com washington.edu
Design an API rate limiter	https://stripe.com/blog/
Design a Stock Exchange (like NASDAQ or Binance)	Jane Street Golang Implementation Go Implemenation
Add a system design question	Contribute

联盟

在下面的文章中，不要把重点放在具体的细节上，而是：

您面试的公司的架构

您遇到的问题可能来自同一个域

Identify shared principles, common technologies, and patterns within these articles
Study what problems are solved by each component, where it works, where it doesn’t
Review the lessons learned

Type	System	Reference(s)
Data processing	MapReduce – Distributed data processing from Google	research.google.com
Data processing	Spark – Distributed data processing from Databricks	slideshare.net
Data processing	Storm – Distributed data processing from Twitter	slideshare.net

Data store	Bigtable – Distributed column-oriented database from Google	harvard.edu
Data store	HBase – Open source implementation of Bigtable	slideshare.net
Data store	Cassandra – Distributed column-oriented database from Facebook	slideshare.net
Data store	DynamoDB – Document-oriented database from Amazon	harvard.edu
Data store	MongoDB – Document-oriented database	slideshare.net
Data store	Spanner – Globally-distributed database from Google	research.google.com
Data store	Memcached – Distributed memory caching system	slideshare.net
Data store	Redis – Distributed memory caching system with persistence and value types	slideshare.net

File system	Google File System (GFS) – Distributed file system	research.google.com
File system	Hadoop File System (HDFS) – Open source implementation of GFS	apache.org

Misc	Chubby – Lock service for loosely-coupled distributed systems from Google	research.google.com
Misc	Dapper – Distributed systems tracing infrastructure	research.google.com
Misc	Kafka – Pub/sub message queue from LinkedIn	slideshare.net
Misc	Zookeeper – Centralized infrastructure and services enabling synchronization	slideshare.net
	Add an architecture	Contribute

切分

Company	Reference(s)
Amazon	Amazon architecture
Cinchcast	Producing 1,500 hours of audio every day
DataSift	Realtime datamining At 120,000 tweets per second
Dropbox	How we’ve scaled Dropbox
ESPN	Operating At 100,000 duh nuh nuhs per second
Google	Google architecture
Instagram	14 million users, terabytes of photos What powers Instagram
Justin.tv	Justin.Tv’s live video broadcasting architecture
Facebook	Scaling memcached at Facebook TAO: Facebook’s distributed data store for the social graph Facebook’s photo storage How Facebook Live Streams To 800,000 Simultaneous Viewers
Flickr	Flickr architecture
Mailbox	From 0 to one million users in 6 weeks
Netflix	A 360 Degree View Of The Entire Netflix Stack Netflix: What Happens When You Press Play?
Pinterest	From 0 To 10s of billions of page views a month 18 million visitors, 10x growth, 12 employees
Playfish	50 million monthly users and growing
PlentyOfFish	PlentyOfFish architecture
Salesforce	How they handle 1.3 billion transactions a day
Stack Overflow	Stack Overflow architecture
TripAdvisor	40M visitors, 200M dynamic page views, 30TB data
Tumblr	15 billion page views a month
Twitter	Making Twitter 10000 percent faster Storing 250 million tweets a day using MySQL 150M active users, 300K QPS, a 22 MB/S firehose Timelines at scale Big and small data at Twitter Operations at Twitter: scaling beyond 100 million users How Twitter Handles 3,000 Images Per Second
Uber	How Uber scales their real-time market platform Lessons Learned From Scaling Uber To 2000 Engineers, 1000 Services, And 8000 Git Repositories
WhatsApp	The WhatsApp architecture Facebook bought for $19 billion
YouTube	YouTube scalability YouTube architecture

反规格化

想要添加博客吗？为避免重复工作，请考虑将您的公司博客添加到以下回购中：

有兴趣添加一个部分或帮助完成一个正在进行的部分吗？贡献自己的力量！

Source(s) and further reading

在整个回购过程中提供积分和来源

kilimchoi/engineering-blogs

联系信息

特别感谢：

Distributed computing with MapReduce
Consistent hashing
Scatter gather
Contribute

许可证

请随时与我联系，讨论任何问题、问题或评论

我的联系信息可以在我的GitHub页面上找到

了解如何设计大型系统

我在开放源码许可下向您提供此存储库中的代码和资源。因为这是我的个人存储库，您获得的我的代码和资源的许可证来自我，而不是我的雇主(Facebook)

动机

向开源社区学习

Anki抽认卡

Copyright 2017 Donne Martin

Creative Commons Attribution 4.0 International License (CC BY 4.0)

http://creativecommons.org/licenses/by/4.0/

声明：本站所有文章，如无特殊说明或标注，均为本站原创发布。任何个人或组织，在未征得本站同意时，禁止复制、盗用、采集、发布本站内容到任何网站、书籍等各类媒体平台。如若本站内容侵犯了原著者的合法权益，可联系我们进行处理。

System-design-primer 学习如何设计大型系统 为系统设计面试做准备

为系统设计面试做准备

编码资源：交互式编码挑战

步骤1：概述用例、约束和假设

系统设计主题索引

步骤2：创建高级设计

学习指导

如何处理系统设计面试问题

系统设计面试问题及解决方案

面向对象的设计面试问题及其解决方案

步骤3：设计核心组件

步骤4：调整设计比例

粗略计算

来源和进一步阅读

Design Pastebin.com(或bit.ly)

设计Twitter时间表和搜索(或Facebook提要和搜索)

系统设计主题：从此处开始

设计一个网络爬虫程序

Design Mint.com

设计社交网络的数据结构

为搜索引擎设计键值存储

按类别功能设计亚马逊的销售排名

设计可扩展到数百万AWS用户的系统

第1步：复习可扩展性视频课程

步骤2：查看可伸缩性文章

性能与可扩展性

延迟与吞吐量

下一步

盖子定理

弱一致性

可用性与一致性

最终一致性

一致性模式

强一致性

可用性模式

故障转移

SQL调优

键值存储

缺点：故障转移

域名系统

复制

数量上的可用性

缺点：DNS

推流CDN

内容交付网络

拉取CDN

文档存储

宽列存储

劣势：CDN

第4层负载均衡

图形数据库

第7层负载均衡

来源和进一步阅读：NoSQL

后备缓存

直写

In sequence

In parallel

负载均衡器

水平缩放

缺点：负载均衡器

反向代理(Web服务器)

负载均衡器与反向代理

缺点：反向代理

微服务

服务发现

应用层

缺点：应用层

关系数据库管理系统(RDBMS)

NoSQL

写后(回写)

SQL或NoSQL

客户端缓存

数据库

CDN缓存

Web服务器缓存

数据库缓存

高速缓存

应用程序缓存

数据库查询级别的高速缓存

对象级别的缓存

System-design-primer 学习如何设计大型系统为系统设计面试做准备