Document upstream RDS ReadIOPS spikes caused by streaming backpressure

[kwannoel]

Context

Slack transcript (no public thread link provided):

• [2026-02-17T03:36:38.751Z] <@U0AF0DZ7C9Y>: <@U03EA02K5L5> /docs file some docs about upstream rds having readiops read spikes because of streaming backpressure.

Related docs/context from search_docs_context:

• https://github.com/risingwavelabs/risingwave-docs/blob/42e1abc46d433e5fc9be2d8ea4d0b2df00001d44/integrations/sources/postgresql-cdc.mdx
• https://github.com/risingwavelabs/risingwave-docs/blob/42e1abc46d433e5fc9be2d8ea4d0b2df00001d44/ingestion/monitor-cdc-progress.mdx
• https://github.com/risingwavelabs/risingwave-docs/blob/42e1abc46d433e5fc9be2d8ea4d0b2df00001d44/deploy/k8s-cluster-scaling.mdx

Problem

Users can observe spikes in upstream AWS RDS ReadIOPS during streaming backpressure, but this symptom/cause path is not clearly documented in CDC troubleshooting guidance.

Without a documented explanation, users may misdiagnose these spikes as source-side issues instead of downstream backpressure and miss the right remediation steps.

Suggested Fix

Add a troubleshooting section for CDC sources (PostgreSQL/RDS) that explains:

1. The likely causal chain: downstream streaming backpressure -> replication lag/catch-up behavior -> upstream ReadIOPS spikes.
2. How to validate the diagnosis with RisingWave and source-side indicators (CDC lag/progress + streaming/backpressure signals).
3. Practical mitigations and what to tune first (remove downstream bottlenecks, scale RW resources, and monitor recovery after backlog drains).

Suggested placement:

• Primary: integrations/sources/postgresql-cdc.mdx
• Cross-reference: ingestion/monitor-cdc-progress.mdx
• Optional link to scaling guidance in deploy/k8s-cluster-scaling.mdx