This article explains pagination patterns for Amazon Aurora DSQL, a serverless distributed SQL database, and how to optimize them for performance and cost.

• OFFSET/LIMIT pagination works but becomes expensive on deep pages in distributed systems
• Keyset (cursor-based) pagination recommended: consistent performance, lower DPU costs, fewer OCC conflicts
• Implement cursors as opaque tokens encoded in API responses for stateless pagination
• Create composite indexes with INCLUDE clauses to enable index-only scans
• Batch processing must respect 3,000-row transaction limit and use short transactions
• Avoid COUNT(*) queries, deep OFFSET pagination, long-lived transactions, non-deterministic sorts
• Use deterministic sort orders with primary key tiebreakers for stable pagination
• Aurora DSQL uses optimistic concurrency control; fast queries reduce conflict windows

Keyset pagination is the recommended approach for Aurora DSQL workloads, delivering consistent performance and predictable costs as data scales.