Cache avalanche, cache breakdown, cache penetration

Cache breakdown

Cache breakdown is when hot data is requested in large quantities , This hot data key Out of date , Resulting in a large number of requests flooding into the database , Causing a surge in database pressure .

Solution ： Get the distributed lock first when updating the cache after the cache expires , The thread that gets the lock can query the database and update the cached data , If the distributed lock acquisition fails , Prove that a thread is querying the database and updating , Try to get the lock again every second for three times , You can return to the status after three times ： Data is being updated , Please try again later .

Cache penetration

Cache penetration refers to accessing data that does not exist in a database or cache , And users are constantly requesting , For example, visit id by -1 Or a very large data . The user at this time is likely to be the attacker , The attack will cause too much pressure on the database .

Solution ： Add verification in the interface layer ,id<=0 Direct interception of data , Data that cannot be fetched in the cache , In the database, there is no access to , Set cache to key-null, Set the cache time smaller , Such as 30 second （ Prevent data from being unavailable under normal circumstances ）, This prevents attackers from using the same id Violent attack .

Cache avalanche

Cache avalanche refers to the simultaneous expiration of batch data in the cache , And the number of queries is huge , Cause excessive pressure on database query and even downtime . Unlike cache breakdown , Cache breakdown is to query a piece of data concurrently , The cache avalanche is that different data has expired , Flooding into the database leads to a sharp increase in database pressure .

Solution ： Cache expiration time is set randomly , Prevent the simultaneous expiration of a large amount of data at the same time .
Never expired data settings , Refresh regularly , When reading data

• There is no cache , Get the distributed lock first, and then query the database , Cache data
• There is a cache and it is not time to refresh , Direct read cache ;
• There is a cache and the refresh time has reached or exceeded , Read cache directly and refresh cache asynchronously , Update cache time ;