I'm unhappy with our current Redshift setup and costs keep growing

I can help with that. Redshift cost growth usually comes from one or more of these buckets:

1. Overprovisioned cluster size

   • Nodes too large or too many nodes for actual workload.
   • Concurrency scaling or resize behavior masking the real need.

2. Inefficient queries / workloads

   • Full table scans, bad joins, missing sort/dist key design.
   • BI tools issuing many small, repetitive queries.

3. Storage bloat

   • Old data not archived or deleted.
   • Too many copies of data across schemas/tables.
   • Vacuum/analyze overhead from frequent updates/deletes.

4. Operational extras

   • Spectrum, data transfer, snapshots, cross-region replication.
   • Keeping dev/test clusters running 24/7.

5. Poor workload isolation

   • ETL + ad hoc + dashboards all on the same cluster causing you to scale up “just in case.”

What I’d recommend

Quick wins

• Review top 20 queries by total runtime and top 20 by bytes scanned.
• Check cluster utilization over 7/30 days:
  • CPU
  • queue time
  • memory pressure
  • disk spill
• Identify idle periods and whether the cluster can be paused or resized.
• Look for tables with frequent vacuum/analyze and heavy bloat.
• Audit snapshots and retention.

Cost control options

• Right-size the cluster, or move to a more flexible setup if usage is spiky.
• Separate workloads:
  • ETL on one cluster or schedule
  • BI/dashboard on another
  • dev/test with automatic start/stop
• Optimize tables and queries:
  • Sort keys
  • Distribution keys
  • Compression encodings
  • Materialized views / aggregates for common patterns
• Archive old data to S3 and query with Spectrum only when needed.
• Consider RA3 / managed storage if you’re not already on it.

If you want the fastest path to savings

I can help you build a Redshift cost reduction plan in one of these formats:

• Checklist for your team
• 90-minute investigation plan
• Architecture review of your current setup
• SQL queries to find expensive tables/queries
• Migration path if you’re considering moving off Redshift

If you want, send me:

• your Redshift node type/count
• whether you’re on RA3 or DC2
• rough daily query volume
• whether ETL and BI share the same cluster
• any CUR / billing symptoms you’ve noticed

…and I’ll help pinpoint the likely causes and next actions.