Shows: the original, I in the Internet giant maintains thousands of sets of cluster, about redis use some pit on the experience summary, hope to bring you some help

Applicable scenario: concurrency, and wide traffic business

Specification: catch content

Reading: explain catch set reason, reading is more important than catch content



Write in front:

Always after the disaster, to remember the importance of disaster;

Always remember, after eaten kui ever reminded,



A basic specification [5]
1. Must be configured to access password

Unscramble: streaking Redis besides convenient by external steal data, also easy to appear on the internal management risk of wrong operation, such as false even cause data to be covered, lost!



2. Must be started with the root

Unscramble: the design of Redis too flexible, this directly make the attacker can remotely through the root run Redis service access to the operating system root access!



3. It is forbidden to Redis as a persistent storage to use

Interpretation: although Redis support AOF and RDB persistence model, but does not record details of each operation time stamp (MySQL binlog will be detailed records execution time), an unwanted operation is unable to accurately roll back!



4. Different business mixed deployment shall be forbidden to use the same set of Redis

Interpretation: (1) the Redis for the single thread model, different business data is stored in together, in addition to management chaos, as long as there is a request under the single thread model command slow, will affect all stored in the all requests with Redis!

(2) although the redis supports multiple db, but requests are not isolated, such as zero library at the request of a slow operation, also a blocks all other 1, 2, 3, 4 repository connection and request!



5. Select a relatively new version, it is strongly recommended that more than 5.0 version

Interpretation: in addition to version 5.0 introduces new features and bug fixes, more memory fragments of Redis 5.0 brought greatly enhance management efficiency, and debris is definitely one of the main performance killer Redis, but early 2 x, 3 x, 4 x version is not good to solve this problem!



? See the end of this article have harvest? Please forward to share with more people

Attention (database architect), promote database skills

Blog: https://blog.csdn.net/samyunhuan/article/details/106151516