Cluster unification

Clusters are such an integral part of Materialize that we want to make sure you think about them as little as possible. 🧠 Over the past few months, we’ve worked on making clusters more flexible and unopinionated about what runs on them (aka cluster unification). For you, as a Materialize user, this means two things:

1. Depending on the characteristics of your workload, you now have the option to manage all types of objects in the same cluster. This gives you more control over capacity planning, and smooths out the experience of onboarding new users to your Materialize organization. We still recommend using separate clusters to manage ingestion and compute — but that’s a feature, not a bug now!

2. Although sources and sinks must be associated with a cluster, if you don’t explicitly specify one using the IN CLUSTER clause on creation, the objects will now be created in the active cluster for the session.

   sql

   copy code

   copied successfully

   --Create the object in a specific cluster
   CREATE SOURCE json_source
   IN CLUSTER some_cluster
   FROM KAFKA CONNECTION kafka_connection (TOPIC 'ch_anges')
   FORMAT JSON;
   
   --Create the object in the active cluster
   CREATE SOURCE json_source
   FROM KAFKA CONNECTION kafka_connection (TOPIC 'ch_anges')
   FORMAT JSON;
   ```

If you’re just getting started, this also means that you no longer have to learn about cluster specifics as the Materialize 101. Off the bat, new Materialize regions include a default cluster named quickstart, which does what it says on the tin: give you resources to just start building.