PostgreSQL 9 High Availability Cookbook

Building a sharding API

When building a horizontally scalable system, we need a database library that facilitates its use. Without this, ad hoc tables can derail the whole process by producing a heterogeneous environment incompatible with a horizontal architecture. We need consistency if we also want reliability.

In the previous recipe, we discussed the necessary components of a function that can generate unique IDs across thousands of logical shards. This will form the core of our API as it ensures that ID collisions are avoided within our application. However, what about the rest? How do we manage each shard? How do we add tables to the application? How can we automate as much management as possible to encourage adhering to the API?

This recipe ...