.webp)
PostgreSQL-based architecture simplifies distributed workflow management and eliminates need for specialized orchestration platforms.
Jeremy Edberg and Qian Li from DBOS presented an unconventional architectural paradigm at QCon San Francisco 2025 positioning PostgreSQL as an orchestration layer rather than merely a data storage system. The approach directly addresses persistent challenges in distributed systems like frequent workflow failures, complex recovery mechanisms and limited visibility into workflow state across enterprise applications. According to Li, "Your database is all you need," highlighting that most organizations already possess sufficient infrastructure to implement workflow support through application level wrapper libraries rather than deploying specialized orchestration platforms.
DBOS Transact, available as open source MIT licensed libraries for Python, TypeScript, Go, and Java, inverts traditional architecture by translating workflows directly into database operations. This database-backed model enables practical operational capabilities addressing everyday challenges. Workflow management becomes possible through standard SQL queries, allowing teams to list, search, cancel, and resume workflows using familiar database operations. The presenters demonstrated a "fork" debugging capability enabling teams to restart workflows from specific steps by copying original inputs and outputs, then replaying execution with updated code—significantly simplifying bug fixes and event replay processes.
The architecture acknowledges inherent challenges. Lock contention emerges when multiple workers access shared queues; the team mitigates this through PostgreSQL's "FOR UPDATE SKIP LOCKED" clause, enabling efficient concurrent processing. Decentralized cron scheduling presents another consideration: rather than maintaining central schedulers, each worker runs identical schedulers using scheduled time as unique workflow identifiers, with random jitter preventing thundering herd problems. Testing workflows becomes more straightforward within this model, as workflows function identically across local development and production environments. Checkpoint mechanisms enable simpler mocking and state management during unit testing.
While Windows Workflow Foundation pioneered SQL Server-based workflow persistence nearly two decades ago, that approach required DSL-based definitions and substantial configuration overhead. DBOS differentiates by utilizing lightweight code annotations in mainstream programming languages, reducing adoption barriers and enabling broader enterprise implementation of database-centric workflow orchestration.
Business Honor views DBOS's database-centric approach as a strategic shift simplifying enterprise workflow orchestration.
.webp)
