Build scalable apps
Early in the quickstart, we created a server group using the basic tier. The basic tier is good for apps that a single node database node (64vcore, 256-GB RAM and 512-GB storage) can handle for the near future (~6 months). Later, you can add more nodes, rebalance your, data and scale out seamlessly.
If your app needs requires multiple database nodes in the short term, start with the Hyperscale (Citus) Standard Tier.
Tip
If you choose the Basic Tier, you can treat Hyperscale (Citus) just like standard PostgreSQL, and achieve full feature parity. You don’t need any distributed data modeling techniques while building your app. If you decide to go that route, you can skip this section.
Three steps for building highly scalable apps
There are three steps involved in building scalable apps with Hyperscale (Citus):
- Classify your application workload. There are use-case where Hyperscale (Citus) shines: multi-tenant SaaS, real-time operational analytics, and high throughput OLTP. Determine whether your app falls into one of these categories.
- Based on the workload, identify the optimal shard key for the distributed tables. Classify your tables as reference, distributed, or local.
- Update the database schema and application queries to make them go fast across nodes.
Next steps
Before you start building a new app, you must first review a little more about the architecture of Hyperscale (Citus).
Feedback
Submit and view feedback for