Skip to main content

​
Overview

Serverless backfilling in RisingWave Cloud runs materialized view backfill jobs on dedicated backfiller resources instead of your main streaming compute nodes. When a serverless backfilling job starts, RisingWave Cloud provisions a temporary resource group for that job by using the configured Backfiller SKU and Replicas as template values. This page focuses on enabling and sizing the Cloud resources. For SQL usage, session variables, WITH clause options, limitations, and monitoring, see Serverless backfilling.

​
Configure serverless backfilling

Serverless backfilling settings are separate from the main cluster setup. To enable it:
  1. Open your project in the RisingWave Cloud console.
  2. Select Resources in the left sidebar.
  3. On the Resources page, scroll to the Backfilling section. The Serverless backfilling row shows its current status and any active backfill jobs. When the feature is off, the status is Disabled and an Enable button is shown.
  1. Click Enable to open the configuration dialog. Set the Backfiller size (SKU) and the number of Replicas for the isolated worker profile used by backfill jobs — the dialog shows the resulting Total backfiller RWUs. Click Enable to confirm.

​
Backfiller SKU

The Backfiller SKU defines the compute size used when RisingWave Cloud provisions the temporary resource group for a serverless backfilling job. Choose a larger SKU for faster backfill throughput on large upstream tables, or a smaller SKU to minimize cost.

​
Replicas

Replicas sets the default replica count used when RisingWave Cloud provisions the temporary resource group for a serverless backfilling job:
  • Increase the replica count if you expect more concurrent backfill work or need higher throughput.
  • Use a smaller replica count if you want to keep backfill resource usage lower.
A large Backfiller SKU or replica count produces high write throughput while a backfill runs, which increases compaction load. If your compactor is sized only for steady-state traffic, it can fall behind and cause compaction back pressure or write stalls. For example, a project that runs steadily on a 4 RWU compactor can stall if you start a 32 RWU serverless backfilling job without scaling compaction.We strongly recommend enabling Serverless Compaction so compactors scale automatically with the backfill load. Otherwise, scale your compactor manually before running heavy backfill jobs.

​
Scheduling behavior

RisingWave Cloud manages the backfill resources automatically:
  1. When a new serverless backfilling job is submitted, the serverless backfilling controller creates a dynamic resource group for that job by using the configured Backfiller SKU and Replicas.
  2. The backfill job runs in that dynamic resource group during materialized view creation.
  3. After the materialized view is created successfully, the materialized view is migrated to the default resource group.

​
Use serverless backfilling in SQL

After enabling serverless backfilling in the Cloud console, use either SET enable_serverless_backfill = true or WITH (cloud.serverless_backfill_enabled = true) when creating a materialized view. See Serverless backfilling for:
  • SQL examples
  • Session-variable and per-statement configuration
  • resource_group limitations
  • Monitoring and progress tracking

​
See also

Serverless backfilling

Learn how to use serverless backfilling in SQL, including the session variable, WITH clause, and monitoring options.

Backfill

Learn how RisingWave initializes materialized views from existing data and the different backfill strategies available.

Manage resources

Manage resource groups and databases in your RisingWave Cloud project.