Redis exploration: cache breakdown, cache avalanche, cache penetration

         In today's Internet Environment , It seems that every system will use cache , Big data is rampant , A lot of data because of the traditional mysql Hard disk query will become a system bottleneck . Therefore, the purpose of introducing cache is to improve the response speed of the system , Increase throughput . Put popular data into the cache to reduce the number of database searches . Since caching matters , When applying caching , Problems encountered and solutions , It becomes a must for programmers 、 The interview will be .

         Basic cache logic used in the system

The first axe ： Cache penetration

         Cache penetration refers to querying data that does not exist at all . Because the cache does not exist , So the request will query the database , Because there is no data , It will not be stored in the cache .

terms of settlement

        1. Cache null ： This is very simple , It's easy to think of , I'll also store the problematic data in the cache , Tell cache , It's problematic , Don't pass it back . Avoid sending requests through the cache to the database .

        2. Use of Blum filter ： Use bloom filter , Put the database data , Map to bloom filter . As the first safety door , Filter invalid requests .

The second axe ： Cache breakdown

         Buffer breakdown refers to a Key The hotspot cache data expires at a certain point in time , Just at this time, there are a lot of requests to query this Key, These requests are sent to the database .

 

         terms of settlement

                 Put hotspot data into cache in advance , And set never expire or use scheduled tasks to refresh cached data and expiration time .

                 Using distributed locks , Ensure that when the cache fails, a large number of concurrent messages will not be sent to the database .

         The difference between cache penetration and cache breakdown

                 The cache penetration request is not in the cache , There is no data in the database , Plus a lot of concurrency , Resulting in system paralysis .

                 The cache breakdown request is not in the cache , But some data in the database .

The third axe ： Cache avalanche

         Cache avalanche refers to key Large area failure at the same time , Cause a large number of requests to be sent to the database . It is a special case of buffer breakdown .

         The common reason is ：1.redis The server is down ;2. A large amount of cached data has the same expiration time , At a certain point in time, thunderstorms .

         terms of settlement

                 Spread the expiration time , Set to fixed time + random number , Reduce the possibility of expiration together .

                 Use locks or message queues , Limit the amount of concurrency .

                 Set up multi level cache , image CPU Third level cache , Layer by layer screening .