1. Redis does not have the concept of replication location like mysql , so when the Slave and the Master are disconnected and reconnected, they will take a full snapshot of the Master , all the data of the Slave will be cleared, and the entire memory table will be re-established, which will cause the Slave to restore data in a special way. Slow, but also brought pressure to the Master .
2. Solve the defects of redis itself through active replication , that is , store the data in redis in multiple copies through the business end or proxy middleware . Middleware gizzard developed by Twitter for replication and partitioning ( https://github.com/twitter/gizzard ) .
Although active replication solves the delay problem of passive replication, it brings data consistency problems.
3. Cache data invalidation strategy?