Solutions to cache penetration: 1. Cache empty data; 2. Use Bloom filters. 2. Solution to the cache avalanche: 1. Set the corresponding hotspot key to never expire; 2. Stagger the expiration time, use random generation for the expiration time, and set the expiration time of the hotspot data longer; 3. Combine multiple caches; 4. Procurement Third-party Redis.
The operating environment of this tutorial: Windows 7 system, Redis version 6, DELL G3 computer.
1. Cache penetration
When the key queried by the user is in It does not exist in redis, and the corresponding ID does not exist in the database. At this time, it is attacked by illegal users. A large number of requests will be directly hit on the db, causing downtime and thus affecting the entire system. This phenomenon is called cache penetration. .
Solution 1: Also cache empty data, such as empty strings, empty objects, empty arrays or lists, the code is as follows
if (list != null && list.size() > 0) { redisOperator.set("subCat:" + rootCatId, JsonUtils.objectToJson(list)); } else { redisOperator.set("subCat:" + rootCatId, JsonUtils.objectToJson(list), 5*60); }
Solution Solution 2: Bloom filter
Bloom filter:
Determine whether an element is in an array, as shown below, using binary to store, The memory occupied is relatively small, 0 represents non-existence, 1 represents existence, and the adding query efficiency is very fast. When a value is saved, an algorithm will be used to save the corresponding value to a certain position on the collection of Bloom filters. There may be multiple keys. When a non-existent key value is passed in, it will be matched with the set. If it does not match, a null will be returned.
Disadvantages:
1, 1% The misjudgment rate, when a key does not exist in the Bloom array, but due to this misjudgment rate, it is judged that the key exists under certain circumstances. When the array is longer, the misjudgment rate is lower, and the shorter the array, the misjudgment rate is lower. The higher the rate
2. When we want to delete a certain key value, the content in our database and redis will be deleted, but it cannot be deleted in the Bloom array because there will be a certain position in the array. If we want to delete a pair of keys, we will change 1 to 0, but all key values will be deleted
3. The code complexity will also increase, because we have to maintain an additional set. When we use redis cluster, Bloom filter should be used in combination with redis
## 2. Redis cache avalanche
Cache avalanche: The data in the cache fails in large batches, and then this use requires a large number of requests. However, because all the keys in redis have failed, all requests will be sent to the db, causing downtimeSolution
1. Set the corresponding hotspot key to never expire[Related recommendations:2. Stagger the expiration time, the expiration time is randomly generated, and the hotspot data The expiration time can be set longer, and non-hot data can be set shorter
3. Combining multiple caches, for example: when a request comes in, you can request redis now, and then request memcache when it does not exist in redis. If it does not exist, go again. Request db
4. Purchase third-party Redis (redis on Alibaba Cloud or Tencent Cloud)
Redis video tutorial]
The above is the detailed content of How to solve redis cache avalanche and penetration. For more information, please follow other related articles on the PHP Chinese website!