Changes / Warehouses & lakehouses / Amazon Redshift
End of supporthigh impact
Amazon Redshift DC2 nodes retired April 23, 2026
DC2 node type feature development stops at release P191, and DC2 clusters retire on April 23, 2026 on the LTM track; customers should move to Serverless or RA3.
Who’s affected
Anyone running Amazon Redshift clusters on DC2 node types. Existing DC2 clusters are on the Long Term Maintenance (LTM) track, cannot change tracks, and must move to Serverless or RA3.
Not affected: Teams already on RA3 or Redshift Serverless, and teams that don't use Redshift.
What it costs
The record gives no prices. The bill depends on what you move to: RA3 node pricing or Serverless usage-based pricing, plus migration effort and any overlap while two environments run. Compare against your current DC2 spend using list prices, not negotiated ones.
When to act
The retirement date of April 23, 2026 has already passed, so any DC2 cluster still running is on borrowed time. Plan a migration with testing and cutover as soon as possible, typically 120+ days of lead time.
Blast radius: Cluster endpoints, JDBC/ODBC connection strings, BI data sources, dbt profiles, ETL/ELT jobs, IAM and network settings, and any reserved capacity commitments tied to DC2.
What should you do? decision tree
- you still have DC2 clusters running and haven't started a migrationInventory every DC2 cluster, its owner, size, and workloads today, and treat migration as urgent since the retirement date has passed.
- your workloads are steady and predictableSize and test an RA3 cluster against a snapshot of the DC2 cluster, then plan a cutover.
- your workloads are spiky, intermittent, or hard to sizePilot Redshift Serverless with representative queries and compare cost to RA3 at list prices.
- you have unused or duplicate DC2 clustersSnapshot and delete them rather than migrating them.
- redshift no longer fits your needs or contractEvaluate Snowflake or Databricks as a broader migration, accepting a longer timeline.
Options
- OptimizeRetire idle or duplicate DC2 clusters first, so you only migrate what is actually used.
- UpgradeMove steady workloads to RA3 nodes from snapshots; keeps the provisioned-cluster model and tooling familiar.
- Re-architectMove variable or bursty workloads to Redshift Serverless and set usage limits to control cost.
- MigrateIf the move is forced anyway, consider moving to Snowflake or Databricks, at the cost of a full migration project.
Alternatives editorial picks, not sponsored
The facts straight from the source
Facts above are extracted from the source and checked word-for-word before publishing. The guidance sections are AI-assisted analysis based only on those facts: a starting point, not legal or procurement advice. Spotted an error? Tell us.
Related changes
- SnowflakeSnowflake: new service users can no longer use passwords (LEGACY_SERVICE blocked)
After enforcement, new non-human users must be TYPE=SERVICE, and LEGACY_SERVICE is invalid in CREATE USER or ALTER USER.
End of support - SnowflakeSnowflake blocks password sign-in for all remaining legacy service users
Non-human users are blocked from password authentication, and existing LEGACY_SERVICE users are migrated to TYPE=SERVICE.
End of support - SAPSAP BW 7.5 mainstream maintenance ends
Mainstream maintenance for SAP BW 7.5 ends December 31, 2027, with extended maintenance available through 2030.
End of support - SAPSAP BW 7.5 extended maintenance ends
Extended maintenance for SAP BW 7.5, BI Java based Broadcasting and BEx Query Designer ends December 31, 2030.
End of support