Buffer pool in MySQL

summary

innodb The storage engine loads data in page units . When we add, delete, and modify a table ,mysql Will find the corresponding table space , Find the segment according to the index , Find the page or area according to the segment , And then 16k The whole page of data is loaded into memory , Then modify the records in the page , Save to disk . Actually, this is not the case , The last step is not to write back to the disk immediately , It's about writing first Buffer Pool in , When the time comes , Will be unified to brush the disk , such mysql Can save some io operation , Especially hot data that is frequently operated . In other words, data modification is temporarily written to memory (Buffer Pool) Of , In fact, there will be persistence problems , Follow up discussion redo,undo The log details how to solve the persistence problem .

Buffer Pool principle

1. To cache pages in memory , When the system starts, a certain amount of continuous memory space will be allocated , This space will 
   Cut it into n and 16k Page of , The control block is not counted in it .
  
2. To distinguish between blank cache pages and modified dirty pages , Two lists are opened. One is free Linked list and flush Linked list 
   When a cache page is occupied by data, it will start from free List removal , At the same time insert into flush Linked list 
  
3. To improve cache hit rate , Have a LUR Linked list , If a page is accessed , Then it will be referred to the head node ,
   such