PostgreSQL 12 High Availability Cookbook - Third Edition

Keeping things fast by creating a cache

In modern applications, it is becoming increasingly common to inject a secondary cache layer between the application and database. This layer stores commonly retrieved data in memory for immediate use. This layer might be composed of memcached or a NoSQL database, such as CouchDB, MongoDB, or Redis.

Once such a layer exists, it's important that the shard_map table is one of the first tables copied there. It has very few rows (a maximum of 2,048 is tiny in the database world), and storing it in memory removes the relatively expensive round-trip to the database. With this mapping in memory, the application will always and immediately know which physical server contains the data it seeks for a specific ...