Construction of module 5 of actual combat Battalion

Design microblog system ” Weibo comments “ High performance and high availability computing architecture

【 Job requirements 】

Based on modules 5 The first 6 Practical cases of microblog in class , analysis “ Weibo comments ” The business features of this core scenario , Then design its high-performance and high availability computing architecture , Including but not limited to the following :

1.  Computational performance prediction ( There is no need to consider storage performance );

2.  High performance computing architecture for non hot events , You need to consider whether to split separate services ;

3.  High availability computing architecture for hot events .

【 Tips 】

1.  Comparison of analytical methods “ Look at micro-blog ” and “ Micro-blog ” The case of .

One 、 Computational performance prediction

background ：2020.9 Live every month 5.11 Billion , Diurnal activity 2.24 Billion ( Reference resources 《 Microblogging 2020 User development report 》).

1.  Comment on

Suppose that the average person sends 1 Micro-blog  &  The average number of people watching a microblog is 100 Time , Among them is 30% Your users will comment , Then the number of comments is :

2.5 Billion  * 100 * 30% = 75 Billion

Let's assume that most people send comments in the morning 8:00~9:00 spot , At noon, 12:00~13:00, evening 20:00~22:00, It is assumed that the proportion of the total number of comments in these time periods is 60%, Then this 4 The average number of hours of tweeting  TPS  The calculation is as follows :

75 Billion  * 60% / (4 * 3600) ≈ 300 K/s

2.  See the comments

Usually, users will load microblog comments at the same time when they watch microblog , So the number of times users read comments is equal to the number of times they read microblogs , The number of comments is ：

2.5 Billion  * 100 = 250 Billion

Suppose that the time period for most people to watch Weibo coincides with the time period for commenting , So the average number of people watching microblog  QPS  The calculation is as follows :

250 Billion  * 60% / (4*3600) = 1000K/s

Two 、 High performance computing architecture for non hot events

1.  Comment on

Business characteristic analysis

Commenting is a typical write operation , So you can't use cache , Load balancing can be used .

The business tolerance of commenting is high , You can handle... Asynchronously , Therefore, a variant of the leaky bucket algorithm can be used -

Write buffer

To achieve .

Architecture analysis

Over 100 million users , You should use a multi-level load balancing architecture , Cover  DNS -> F5 -> Nginx ->  Multilevel load balancing of gateway .

Architecture design

1） Load balancing algorithm selection

Users can send requests to any server when making comments , Choose here “ polling ” perhaps “ Random ” Algorithm .

2） Estimate the number of business servers

Commenting like microblogging involves several key processes ：

• Content review ( Rely on the audit system )、 Write data to storage ( Dependent on storage system )、 Data write cache ( Rely on the cache system )

• Suppose an asynchronous write user can wait 30 second

• Per service per second 500 To estimate

complete  300K/s  Of  TPS, need  

300K/500/30

 = 20 Servers , Plus a certain amount of reserve ,25 One server is enough .

2.  See the comments

Business characteristic analysis

Reading comments is a reading scenario and comments cannot be modified , Therefore, it is very suitable to use cache architecture , At the same time, due to the large number of requests , Load balancing architecture also needs .

Architecture analysis

1） Over 100 million users , You should use a multi-level load balancing architecture .

2） Multi level cache architecture can be used for high requests .

Architecture design

1） Load balancing algorithm selection

Anyone can read the comments , So choose “ polling ” perhaps “ Random ” Algorithm

2） Estimate the number of business servers

hypothesis  CDN  Capable of carrying 90% Of users , So the rest 10% Your request to read comments will enter the system . request  QPS  by  

1000K/s * 10% = 100K/s

. Because the processing logic of read operation is relatively simple , Mainly read cache system , Therefore, it is assumed that the processing capacity of a single business server is 1000/s, Then the number of machines is 100 platform , according to 20% Reserved quantity of , The final number of machines is ：

100K / 1000 * (1 + 20%) = 120 platform

Overall architecture design

3、 ... and 、 High availability computing architecture for hot events

Compared with microblog, microblog comments on hot events have lower requirements for high availability , We can consider limiting the flow of comments on the microblog of hot events , Degrade and fuse if necessary .

Writing comments is highly available

Current limiting ： The asynchronous write buffer scheme is used for traffic peak shaving , As shown in the figure “ Write comment asynchronous write buffer scheme ”. In the above normal comment writing design, an infinite leaky bucket has been used for buffering , The current limit here can not consider the superposition scheme more

View comments highly available

Multi copy cache architecture can be adopted to deal with hot spots . When the replica cache server is under too much pressure, it will automatically blow .