Weibo comments on high availability and high performance computing architecture

1  Computing performance evaluation

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

Comment on

Considering that microblog is a business that sees more and sends less , Suppose that the average person sends 4 comments （ Only consider text microblog ）, The daily sending volume of microblog is about  10 Billion bars .  Most people tweet in the morning 8：00~9：00 spot , At noon, 12：00~13：00, evening 20：00~22：00, Suppose you send microblogs in these time periods   The proportion of the total amount is 60%, Then this 4 The average number of hours of tweeting  TPS  The calculation is as follows ： 10 Billion  * 60% / (4 *3600）= 41.6K/S

See the comments

Because the vast majority of commenting users view comments on large V And stars , Therefore, we assume that the average number of people watching a microblog is 100 Time , The number of times you watch microblog is ： 2.5 Billion  * 100 = 250 Billion .  Most people's time period of watching microblog basically coincides with that of sending microblog , So the average number of people watching microblog  QPS  The calculation is as follows ： 250 Billion  * 60% / (4*3600) = 1000 K/S

2  Non hot event high performance computing architecture

Comment on

【 Business characteristic analysis 】 

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

【 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   Depending on login status when posting comments , The login status is generally saved in the distributed cache , So when commenting , Send the request to any server   All servers are OK , Choose here “ polling ” perhaps “ Random ” Algorithm . 

2.  Estimate the number of business servers   Commenting involves several key processes ： Content review （ Rely on the audit system ）、 Write data to storage （ Dependent on storage system ）、 Data write cache （ rely on   Cache system ）, So, per service per second 500 To estimate , complete 41.6K/s Of  TPS, need 83 Servers , Plus a certain amount of reserve ,90  This server is almost

See the comments

【 Business characteristic analysis 】 

Reading comments is a typical reading scene , Because the comments can't be modified after they are sent , Therefore, it is very suitable to use cache architecture , At the same time, due to the large number of requests , Load balancing rack   Structure also needs . 

【 Architecture analysis 】

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

 2.  The requested quantity reaches 250 Billion , You should use a multi-level cache architecture , In especial  CDN  cache , Is the core of cache design . 

【 Architecture design 】 

1. Load balancing algorithm selection   Visitors can read the comments directly , Therefore, sending a request to any server can , Choose here “ polling ” perhaps “ Random ” Algorithm .

 2.  Estimate the number of business servers   hypothesis  CDN  Capable of carrying 90% Of users , So the rest 10% The request to read comments enters the system , Ask for  QPS  by 1000K/s * 10% = 100K/s,  Because the processing logic of reading comments 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 120 platform

3  Non hot event high availability computing architecture