Cache penetration, cache breakdown, cache avalanche and Solutions

redis It's a kind of NoSql High speed slow K-V Save the database , Can achieve fast access .

redis More suitable for storage Session Information 、 cache 、 Ranking List 、 Message queue 、 Friends and counters

redis Cache penetration

redis Cache penetration refers to querying data that must not exist , But there is no such data in the cache , You need to query from the database , And there is no such data in the database , So it will not be written to redis in , This leads to querying the database every time you access the nonexistent data , This leads to cache penetration .

terms of settlement ：

• Cache null values , If the data cannot be queried from the database , We can set a null value with a short expiration time to be stored in redis in

• Using interceptors , Judge the accessed data , Shield the data that must not exist .

• Use of Blum filter , Use a larger bitmap To filter key.

The bloon filter

The working principle of Bloom filter is well understood , Is to store objects in a hash table , To judge whether the object exists . In order to improve judgment , Multiple hash functions can be used for hash distribution .

Collective implementation method of Bloom filter , Pass an object through multiple hash functions , Get different hash values , And then bitmap The position in is determined by 0 Set as 1, For example, one Tom There may be three hash functions , Exist in sixteen spaces bitmap in , Location may be 3,5,15.

The reason why objects need to be hashed many times is , Different objects , After the same hash function , The hash value may be the same , That is to say Hash collision , So we need to try to reduce this possibility .

So to sum up , If we want to judge whether an object exists , You need to pass the object through the hash function in turn , Calculate the hash value , As long as bitmap There is a position in 0, Then the object must not exist . If all positions of the object are 1, Then the object may exist （ Hash collision may cause us to misjudge ）.

redis Cache breakdown

Cache breakdown is a special case of cache avalanche , It refers to a hot data , At the moment the cache expires , There are a lot of requests to access this data , Cause breakdown to DB, cause DB Too much pressure on me .

resolvent

• Use synchronized+ Double check mechanism , Take data from DB Is written to the cache , reduce DB pressure .

• Never expired data settings

Redis Cache avalanche

Cache avalanche means that the cache concentration disappears within a period of time , Cause a lot of cache breakdown , thus DB Too much pressure load .

resolvent

• Use synchronized+ Double check mechanism , Control the number of reads from the database and writes to the cache by locking , Some key Only one thread is allowed to query and write to the cache , Other threads wait .

• Add a random value to the cache expiration time , Reduces cache expiration at the same time .

• Cache preheating , Avoid projects starting , The cache is empty , Cause a lot of requests to DB, Load the data into the cache in advance .