SQL Server to PostgreSQL, without the risky weekend
Moving to PostgreSQL removes SQL Server licensing and gives you an open-source database that runs anywhere. The migration itself should be a planned engineering project, not a leap of faith.
Why migrations go wrong
The data copy is rarely the hard part. Problems come from what quietly depends on the database: stored procedures, scheduled jobs, reports, integrations, and behaviour differences such as case-sensitive text comparison and timestamp handling. A migration plan has to find all of that before cut-over day.
What's included
Assessment
Inventory of objects, consumers and SQL Server-specific features, with a risk list and effort estimate.
Schema & type mapping
Deliberate choices for identity columns, collation, timestamps, money and binary types.
Code conversion
Stored procedures and functions rewritten in PL/pgSQL, with review rather than blind conversion.
Data validation
Row counts, checksums and business-level checks that prove the data arrived intact.
Performance testing
Real workloads replayed, query plans compared, indexes tuned for PostgreSQL.
Cut-over
A written runbook, go/no-go checkpoints and a rehearsed rollback path.
Further reading
A detailed walk-through of the approach: SQL Server to PostgreSQL migration: a practical checklist.
Questions
How long does a SQL Server to PostgreSQL migration take?
It depends on database size, the amount of T-SQL code and how many applications connect to it. The assessment phase produces a realistic timeline before any commitment.
Can the migration happen without downtime?
Near-zero downtime is possible using change data capture, so only the final changes sync at cut-over. Whether it is worth the extra effort depends on the system.
Do our applications need to change?
Usually some. Connection strings, drivers and any SQL written directly in application code need review. This is identified during assessment.
Can you migrate to a managed cloud database?
Yes, including Amazon RDS or Aurora for PostgreSQL and Google Cloud SQL.