Redis persistence

One 、 Persistence concept

Persistence has a wide range , From memory to hard disk, this is called persistence ; Data backup is also called persistence （ therefore mysql It's also persistent ）

Two 、redis Persistent classification

1、RDB： Take a snapshot

Update the data every time point . The occupied volume is relatively small 、 Fast recovery ; But the data loss may be great （ Default RDB, It can be turned on manually AOF, Once on AOF You will no longer read snapshot files, but only log files ）

2、AOF： journal

Close to real-time updates （ This is configurable ）. The occupied volume is relatively large 、 Slow recovery ; But data loss may be small

AOF Two modes ：

1. Real time updates ： Write and update the log . because redis It is a single thread. If it is completely real-time, the efficiency will be very low , Every time you write data to memory, you have to write it to hard disk , Then you can process other requests
2. hand os： The operating system is based on pagecache（ Page caching ） To update , every other 5s Update and reach 10% memory update , These two work together . Data will be lost but no RDB serious ）

Because it's a log , So every update operation will be recorded , But the last record is actually useful , So the volume is big 、 Slow recovery . But you can configure rewriting to update useless information , However, this is not real-time, but there are trigger conditions, such as extremely large or to a certain point in time . here we are 4.x After the version AOF Improved , When it comes to rewriting , It does not loop through and delete useless information , It takes snapshots directly in memory , Put the snapshot directly into AOF Replace the original record （ Delete the original directly ）, Then new