Although the memory service has been used, why the cache service is used still?

Under the conditions with the same data requests, the response speed of the cache service is much higher than the memory service, but its capacity is relatively smaller and the price is relatively higher than the memory service. The cache service is generally used to accelerate the access to hot data.

To use the cloud database SCS, how much the original system i modified?

The cloud database SCS is compatible with Memcache and Redis of native APIs respectively to minimize the code modifications.

Does the cloud database SCS for Redis support CPU selection?

The single-node CPU processing capacity of the cloud database SCS is based on a single core. The larger the number of multiparts is, the higher the number of CPU cores is.

The cloud database SCS for Redis does not support "select" command and cannot be fragmented. How to solve this problem?

Currently, the Redis architecture in SCS provided by Baidu AI Cloud is different from the bare Redis clusters. The cloud database SCS builds a Proxy external service on many Redis instances for automatic hashing the multiparts for users. The bottom layer is equivalent to multiple bare Redis instances. In fact, each Redis instance is a multipart; therefore, the original select command is not very meaningful.

What restrictions and precautions should be noted when the cloud database SCS for Redis instance capacity is changed?

• Single instance: The cluster cannot be written and can be read only when the capacity is scaled. Users with long linkage should reestablish the connection after the capacity is scaled.
• Cluster: The key under being migrated during the capacity scaling cannot be written.

Should the cloud database SCS be enabled for persistence?

The cloud database SBC provides “memory and disk” for data storage. The default is to persist to the disk by use of aof. If there is incremental data, the disk is rewritten once every second, and aof-rewrite is triggered once the data growth reaches 100%, which results in a certain delay in processing of the user requests. For some pure cache users, the persistence to disk is not necessary, in which case you can deactivate the disk persistence.