1, the query:
After the direct return unique index found to meet the conditions of data
Whether it will continue to match the data found in the general index to meet the conditions to meet the conditions of the next line
Impact: data will be loaded into memory by page, line overhead and more unlikely unless matched exactly match the last data of the current page, we need to continue to load a page from disk
2, write
The only index: need to determine whether to repeat, you need the corresponding page loaded from disk memory
Ordinary Index: General Index to use change buffer (space buffer pool), the update directly written into buffer, do not read the disk
3、change buffer vs redo log
change buffer: write directly to memory changes, without taking out the page data corresponding to the reduced consumption of random read from the memory
redo log: a sequential write directly to disk when updating to avoid random write performance overhead
4, redo log buffer can cause mysql occasionally shake
After --- redo log filled flush dirty pages need
--buffer need to flush dirty pages shortage